Evaluation of health-related quality of life in individuals with vestibular disease using disease-specific and general outcome measures.

Key Words: Health-related quality of life, Outcome measures, Psychometric properties. Vestibular disease.

Dizziness accounts for 8 million primary care visits to physicians in the United States annually.[1] Vestibular system disorders are the cause of dizziness in approximately 40% to 50% of patients referred to otolaryngologists[2] and to primary care clinics.[3,4] Individuals with dizziness report a variety of symptoms, including vertigo, nausea, postural instability, blurred vision, and disorientation.[5] These symptoms can result in a variety of emotional and physical problems, including emotional distress, anxiety,[6-8] and an inability to perform activities of daily living or work.[9,10] Vestibular system disorders appear to affect health-related quality of life.

Vestibular rehabilitation includes exercises that are used to manage the symptoms of vertigo and imbalance associated with vestibular disorders. The exercises are designed to improve the individual's quality of life. Exercises may include repetitive eye, head, and body movements to reduce motion intolerance and gaze instability and balance activities to reduce postural instability.[11,12] An individual's progress or lack of progress in vestibular rehabilitation is usually measured by observing changes in motion intolerance, balance, functional abilities, and, more recently, health-related quality of life.[13,15]

The Dizziness Handicap Inventory[16] (DHI) and the Medical Outcomes Study (MOS) 36-Item Short-Form Health Survey[17] (SF-36) are two commonly used health-related quality-of-life survey instruments. The DHI, a disease-specific questionnaire, was developed for individuals with dizziness or balance problems and measures how vertigo and disequilibrium (imbalance) affect an individual's quality of life.[16] Disease-specific questionnaires are thought by some experts to be more responsive to clinical change than generic questionnaires.[17] Generic questionnaires, such as the SF-36, use a global approach to measure health status as it relates to an individual's functional well-being.[18] Generic questionnaires may be used with several populations[17,19] and are less responsive to change than disease-specific questionnaires.[19] Generic questionnaires allow comparisons of health status between disabled and nondisabled individuals as well as comparisons among persons with dissimilar medical conditions. Some researchers[20,21] have found parts of the generic questionnaires to be equally as responsive or more responsive to change than disease-specific questionnaires. Patrick et al[21] found the physical function scale of the SF-36 to be more responsive than either the disease-specific Roland Disability Questionnaire or the Sciatic Frequency Index. Kantz et al[20] reported that the physical function scale of the SF-36 was equally responsive as a disease-specific questionnaire for patients with total knee replacements. The more generic SF-36 was developed for general population surveys, clinical practice, research, and health policy assessment.[22] Use of this tool to assess patients with vestibular disease has not been described in the literature. Thus, the health status of individuals with vestibular disease has not previously been compared with the health status of individuals from the general population.

Whether both the DHI and the SF-36 are highly correlated when assessing health-related quality of life in patients with vestibular disorders remains unknown. Whether the DHI and the SF-36 are equally reliable and responsive is also not known. If the questionnaires are highly correlated and are equally reliable and responsive, only one questionnaire may be necessary in clinical practice and research. By understanding the relationship between these questionnaires, practitioners can decide whether either or both questionnaires should be used to assess patients' health status.

Psychometric properties of health-related quality-of-life questionnaires, including reliability and responsiveness, are important in the patient populations for which they are used.[23-25] Reliability of the DHI for patients with vestibular dysfunction was high (r=.97) in a study (N=14) by Jacobson et al,[16] but systematic variation was not assessed. The reliability of the SF-36 has been studied extensively,[26,27] but not for individuals with vestibular dysfunction. More importantly, the reliability of the DHI and the SF-36 has not been estimated concurrently for persons with vestibular dysfunction.

Responsiveness is the ability to detect the minimal clinical change in a variable over time.[28] Responsiveness of an instrument may be measured by following a disease through its natural progression or by measuring change following clinical intervention. Measuring change in health-related quality of life of patients with vestibular dysfunction using the DHI and the SF-36 provides an opportunity to compare the responsiveness of these two instruments.

The purposes of this study were (1) to describe the relationship between the DHI and the SF-36 for individuals with vestibular dysfunction, (2) to estimate the reliability and responsiveness of the DHI and the SF-36 for individuals participating in a vestibular rehabilitation program, and (3) to compare the health-related quality of life of individuals before and after vestibular rehabilitation (SF-36, DHI) and to contrast their health status with that of the general population (SF-36).

Method

Subjects

One hundred eight patients with vestibular dysfunction who were referred to the Vestibular Rehabilitation Clinic, University of Iowa Hospitals and Clinics, Iowa City, Iowa, were invited to complete the DHI and the SF-36 over a 15-month period. Five patients (2 men, 3 women) did not participate in the study due to their inability to complete questionnaires independently. Eight additional patients were eliminated because the DHI was incomplete. The patients provided informed consent, were able to read, and met an otolaryngologist's diagnostic criteria for vestibular dysfunction. The otolaryngologist classified patients into five general groups, which were adapted from Baloh.[5] The five general groups were (1) unilateral vestibular dysfunction, (2) bilateral vestibular dysfunction, (3) central vestibular dysfunction, (4) mixed central and peripheral dysfunction, and (5) nonspecific vestibular dysfunction. The distributions of patients by general categories are presented in Table 1.

Table 1.
General Vestibular Diagnostic Classifications for Patient Subsets

                                            Study Phase

                                            Reliability
General Vestibular Diagnostic Dysfunction   (n=20)

Unilateral dysfunction                      17 (87%)
Bilateral dysfunction                        1 (5%)
Central dysfunction                          0 (0%)
Mixed (central and peripheral)
dysfunction                                  2 (10%)
Nonspecific dysfunction                      0 (0%)

                                            Study Phase

                                            Correlation
General Vestibular Diagnostic Dysfunction   (N=95)

Unilateral dysfunction                      78 (82.1%)
Bilateral dysfunction                        3 (3.2%)
Central dysfunction                          4 (4.2%)
Mixed (central and peripheral)
dysfunction                                  4 (4.2%)
Nonspecific dysfunction                      6 (6.3%)

                                            Study Phase

                                            Responsiveness
General Vestibular Diagnostic Dysfunction   (n=31)

Unilateral dysfunction                      27 (87.1%)
Bilateral dysfunction                        1 (3.2%)
Central dysfunction                          1 (3.2%)
Mixed (central and peripheral)
dysfunction                                  1 (3.2%)
Nonspecific dysfunction                      1 (3.2%)

To estimate reliability, 24 consecutive patients, a subset of the total sample, were tested with both the SF-36 and the DHI. The patients were asked to complete both questionnaires twice, before their physical therapy evaluation and again within a 24- to 48-hour period. Four patients (3 men, 1 woman) did not complete the questionnaires a second time, leaving 20 patients for the final analysis. To describe the relationship between the SF-36 and the DHI, 95 patients completed both questionnaires at their initial physical therapy evaluation. Responsiveness was estimated using 31 consecutive patients, also a subset of the total sample, who received both the SF-36 and the DHI at two time periods. There were no differences in mean age, height, weight, and length of illness among the three groups. Because the 31 patients in the responsiveness group were a subset of the total sample, additional analysis showed that their demographics were not different from the demographics of the other 64 patients. The diagnoses and demographics for the participating patients are shown in Tables 1 and 2. Sixty patients (67%) were female, and 35 patients (33%) were male.

Table 2.
Patient Demographics

                             Study Phase

                             Reliability (n=20)
Variable                      X       SD     Range

Age (y)                     56.45   13.93    36-78
Height (cm)                166.32   10.83   138-183
Weight (kg)                 78.65   19.63    54-124
Length of illness(a)(mo)    27.2    35.7      3-120

                             Correlation (N=95)
Variable                      X       SD     Range

Age (y)                     56.99   14.87    25-88
Height (cm)                167.42    9.56   135-190
Weight (kg)                 82.28   19.34    46-144
Length of illness(a)(mo)    24.83     2.6     1-120

                              Responsiveness (n=31)
Variable                      X       SD     Range

Age (y)                     57.77   15.42    34-88
Height (cm)                168.75    8.95   155-190
Weight (kg)                 78.08   18.75    49-129
Length of illness(a)(mo)     26.2   37.4      3-120

(a) Length of illness: Any length of illness greater than 10 years was collapsed to 120 months.

Instruments

The DHI is a disease-specific questionnaire that was developed to quantify the impact of dizziness on quality of life (Appendix 1).[16] A total possible score of 100 is the result of 36 points from the emotional scale (9 items), 36 points from the functional scale (9 items), and 28 points from the physical scale (7 items). Each question provides a choice of three responses: yes (4 points), sometimes (2 points), or no (0 points). To facilitate the use of the three DHI scales separately in all calculations, each scale's score was calculated as the percentage of the total possible points for each scale. For example, 24 points on the emotional scale received a score of 67% (24/36x100). Next, the total score and the three scale scores were transformed by subtracting each score from 100 so that 0 represented poor function due to dizziness or unsteadiness and 100 represented no problems due to dizziness or unsteadiness. The DHI does not focus its questions on any particular time interval.

The SF-36, which consists of eight health scales, was used to measure generic health-related quality of life (Appendix 2).27 The eight health scales of the SF-36 are (1) physical function (10 items), (2) role limitation due to physical problems (4 items), (3) bodily pain (2 items), (4) general health (5 items), (5) social function (2 items), (6) vitality/energy (4 items), (7) role limitation due to emotional problems (4 items), and (8) mental health (5 items). The standard SF-36 form asks questions regarding the past 4 weeks.

The eight health concepts of the SF-36 are summarized in two summary scales: the physical component summary scale (PCS-36) and the mental component summary scale (MCS-36). Although both scales are calculated from weighted aggregates of all eight SF-36 domains, the PCS-36 and the MCS-36 are heavily weighted with respective SF-36 scores. The PCS-36 is weighted more heavily with physical function, bodily pain, and role limitation (physical) scores. The MCS-36 is weighted more heavily with mental health, role limitation (emotional), social function, and vitality/energy scores.[27]

The SF-36 data were analyzed using a Statistical Analysis Software(*) code developed by the Medical Outcomes Trust at the New England Medical Center.[27,29] For missing data, a computer algorithm substituted the average score of one scale if greater than 50% of the items were present.[27] The eight scales were scored individually and transformed, resulting in a scale ranging from 0 to 100, with 0 denoting poor health. The SF-36 summary scores were calculated according to methods described by Ware et al.[29] Specifically, each of the eight SF-36 scores was standardized using a Z transformation. The mean score of the US population was subtracted from the score of the respective SF-36 scale, then the difference was divided by the respective scale's standard deviation. In the next step, aggregate scores were calculated for the PCS-36 and the MCS-36 using weights for each SF-36 scale determined by factor analysis. Finally, the PCS-36 and the MCS-36 received a linear T-score transformation by multiplying each score by the standard deviation of the general US population ([+ or -] 10) and adding 50 (mean score of the general US population). That is, the PCS-36 and MCS-36 scores were referenced to normative values.

Procedure

Three different procedures were completed for estimating the test-retest reliability of the SF-36 and the DHI, measuring the relationship between the SF-36 and the DHI, and comparing the responsiveness of the SF-36 and the DHI. To estimate the test-retest reliability of the SF-36 and the DHI, 20 patients completed both questionnaires twice over a 24- to 48-hour period. This short time was chosen due to the unpredictable nature of symptoms in patients with vestibular problems.[7,30] The patients completed the two surveys before their physical therapy evaluation and once at home 24 to 48 hours later. The patients completed the SF-36 first, followed by the DHI, for both occasions. The SF-36 survey was completed first to allow the patients to focus first on their general quality of life, and then to focus on the disease-specific factors affecting their quality of life. The patients were contacted by the investigator (LJE) between 24 and 48 hours after their physical therapy appointment to verify completion of the surveys. If a patient had not completed the surveys, he or she was asked to complete them during the telephone call. Two patients completed the surveys during the telephone interview by reading their responses to the investigator.

The relationship between the SF-36 and the DHI was determined with the 95 patients who completed the SF-36 and the DHI prior to their initial physical therapy evaluation. To determine the responsiveness of the SF-36 and the DHI, a subset of 31 patients completed both questionnaires twice. The patients completed the SF-36 and the DHI once at their initial physical therapy visit and once at follow-up to physical therapy ([bar] X=6.9 weeks, SD=2.7). Six to eight weeks was chosen to assess responsiveness because this time was associated with improvement in patients receiving vestibular rehabilitation.[13,14]

Data Analysis

Descriptive statistics for the DHI scores and the SF-36 scores were calculated for the patients in each group. The percentage floor and ceiling effects for both the DHI and the SF-36 were computed for the total sample. Floor effects are present when individuals score poorly, at the bottom of the scale, so that no further decrement may occur. The opposite is true for ceiling effects.

The relationship between the SF-36 and the DHI was described using the Pearson product-moment correlation coefficient (r). Correlations between the scales within each questionnaire were also computed. The test-retest reliability of the SF-36 and the DHI was estimated using the intraclass correlation coefficient (ICC[2,1]) and the Pearson product-moment correlation coefficient. Test-retest reliability was determined for the total DHI score, the three DHI scales, the eight SF-36 scales, and the SF-36 summary scales. The standard error of the measurement (SEM) for the 95% confidence interval was also calculated.

Dependent t tests were completed to determine whether the DHI and the SF-36 changed for the subset of 31 patients before and after 6 to 8 weeks of vestibular rehabilitation. Responsiveness was determined by calculating the mean change as a percentage of the full scale range of all DHI and SF-36 scales over the 6- to 8-week period. The mean change as a percentage of the full scale range was equal to the absolute change of all DHI and SF-36 variables because the possible scores range from 0 to 100 for each questionnaire. For example, a change of 10 points on either questionnaire represents a 10% change with respect to the range (0-100) of the total scale. Unlike the scoring of the DHI and the SF-36 scales, however, the SF-36 component summary scores (PCS-36 and MCS-36 scores) do not range from 0 to 100. Therefore, the SF-36 component summary score range was determined by inputting maximum and minimum values into the formulas for calculating the summary scores.(29) The resulting ranges of 74.55 (1.73-76.28) and 70.73 (3.11--76.28) were determined for the PCS-36 and MCS-36 scores, respectively. Normalizing the SF-36 component summary scores to the total range of the scale enabled comparisons between the DHI summary score and the SF-36 component summary score.

Dependent t tests were used to compare the mean change (as a percentage of full scale range) of the SF-36 scales and DHI scales related to physical health. Comparisons were also made between the mean change of the SF-36 scales and DHI scales related to mental health. Dependent t tests were used based on the assumption that both the SF-36 and the DHI are health-related quality-of-life measures and thus measure similar concepts. Bonferroni adjustments were made for multiple comparisons.

The responsiveness of the DHI and SF-36 scores was also estimated using Guyatt's statistic.[28] Guyatt's responsiveness formula is: clinical change/square root (2 x mean square error). Deyo et al[23] have shown that the denominator of this equation is equivalent to the standard deviation of the score changes across a 24-hour test-retest period; therefore, this calculation was used for the denominator. The values for the denominator were taken from the baseline assessments of the 20 patients participating in the reliability component of the study. The numerator represented the change for the DHI and SF-36 scores and was obtained before and after 6 to 8 weeks of vestibular rehabilitation. The number of patients required to detect this change was calculated according to the procedure described by Guyatt et al,[28] using an alpha level of .05 and beta level of .10 and assuming that independent groups would be selected in a randomized clinical trial.

Results

Completion rates for the SF-36 and the DHI were similar to previously reported completion rates for the SF-36.26 Ninety-five percent (n=98) of the SF-36 questionnaires were complete. Three percent, 2%, and 1% of the patients did not complete 1, 2, and 4 items, respectively. Ninety-two percent (n=95) of the DHI questionnaires were complete, with 8% of the patients missing 1 item.

Descriptive statistics, including the percentage of scores that could not go lower (floor) and the percentage of scores that could not go higher (ceiling), for the DHI and SF-36 scores of the 95 patients are presented in Table 3. The scales with the greatest floor effects were the SF-36 scales role limitation (physical) and role limitation (emotional). The scales with the greatest ceiling effect were the SF-36 scales bodily pain, social function, and role limitation (emotional). The DHI scales exhibited minimal floor and ceiling effects.

Table 3.
Dizziness Handicap Inventory[16] (DHI) and Medical Outcomes Study
36-item Short-Form Health Survey[27,29] (SF-36) Scores (N=95)

Variable                         X       SD     Minimum   Maximum

DHI
  Total                        53.57    20.82    6.00      96.00
  Functional                   51.11    26.01    0        100.00
  Physical                     45.34    23.32    0        100.00
  Emotional                    62.45    21.45    5.56     100.00

SF-36
  Physical function            51.59    26.17    0        100.00
  Role limitation (physical)   22.10    31.76    0        100.00
  Bodily pain                  57.37    28.97    0        100.00
  General health               58.07    21.95    0         97.00
  Vitality/energy              37.79    23.55    0        100.00
  Social function              55.92    28.88    0        100.00
  Role limitation (emotional)  45.61    38.91    0        100.00
  Mental health                65.77    18.51   16.00      96.00
  Physical component summary   36.53    10.25   16.30      56.20
  Mental component summary     43.24    10.42   23.03      64.55

                                 Percentage of    Percentage of
Variable                         Floor Effects    Ceiling Effects

DHI
  Total                          0                  0
  Functional                     4.20               1.00
  Physical                       5.30               1.00
  Emotional                      0                  2.10

SF-36
  Physical function              1.00               2.10
  Role limitation (physical)    55.80               8.40
  Bodily pain                    3.20              18.90
  General health                 1.00               1.00
  Vitality/energy                4.20               1.00
  Social function                2.10              14.70
  Role limitation (emotional)   30.50              25.30
  Mental health                  1.00               0
  Physical component summary     0                  0
  Mental component summary       0                  0

The correlation matrix for the DHI and SF-36 scores is shown in Table 4. The reported probability values signify that the Pearson product-moment correlations were significantly different from zero. The correlations between the DHI and the SF-36 ranged from .11 to .71. The highest correlation was between the social function score of the SF-36 and the DHI emotional score (r=.71). The three scales of the SF-36 that were most related to mental health (role limitation due to emotional problems, mental health, vitality/energy) had correlations between .40 and .50 with the DHI emotional score. The correlation between the PCS-36 score and the DHI functional score was .54. The MCS-36 score and the DHI emotional score also had a correlation of .54.

[TABULAR DATA 4 NOT REPRODUCIBLE IN ASCII]

The test-retest correlation coefficients and the SEMs for the DHI and the SF-36 for the 20 patients in the reliability component of the study are shown in Table 5. The ICCs ranged from .79 to .95 for the DHI and from .64 to .92 for the SF-36. For the DHI, the emotional scale had the lowest SEM. For the SF-36, the PCS-36 and MCS-36 had the lowest SEMs and the role limitation (physical) and role limitation (emotional) scales had the highest SEMs.

Table 5. Dizziness Handicap Inventory[16] (DHI) and Medical Outcomes Study 36-item Short-Form Health Survey[27,29] (SF-36) Test-Retest Reliability Correlation Coefficients

                               Test              Retest
Variable                   X         SD        X         SD

DHI
 Total                   47.70     23.30     52.20      25.54
 Functional              43.61     29.25     47.78      27.71
 Emotional               59.44     24.19     62.78      25.81
 Physical                37.86     24.32     44.29      28.44

SF-36
 Physical
 function                48.81     28.06     54.25      28.20
 Role limitation
   (physical)            25.00     40.56     25.00      38.28
 Bodily pain             63.70     28.43     66.75      27.88
 General health          57.45     21.10     61.70      23.38
 Vitality/energy         46.00     23.93     48.25      24.13
 Social function         54.37     31.49     55.00      32.55
 Role limitation
   (emotional)           48.33     48.33     36.37      41.75
 Mental health           68.10     68.10     69.00      22.62
 Physical component
   summary               36.76     10.12     38.92      8.93
 Mental component
   summary               44.88     12.40     43.35      11.01

Variable                   ICC(a)(2,1)    Pearson r  +/- 2xSEM(b)

DHI
 Total                         .94           .96         9.32
 Functional                    .95           .94        11.70
 Emotional                     .95           .95         9.68
 Physical                      .79           .82        20.64

SF-36
 Physical
 function                      .90           .91        16.84
 Role limitation
   (physical)                  .64           .64        48.67
 Bodily pain                   .90           .90        17.98
 General health                .79           .79        21.16
 Vitality/energy               .91           .91        14.36
 Social function               .90           .90        19.92
 Role limitation
   (emotional)                 .81           .83        36.20
 Mental health                 .91           .92        11.96
 Physical component
   summary                     .85           .87         7.28
 Mental component
   summary                     .92           .94         6.06

(a) ICC=intraclass correlation coefficient.

(b) [+/-] SEM = standard error of the measurement for 95% confidence interval = 2X [(SD) x (square root (1-r)] Test SD was used for SEM calculation.

Scores for all DHI scales and for the PCS-36 and the physical function, role limitation (physical), bodily pain, vitality/energy, social function scales of the SF-36 were changed after an average of 7 weeks of rehabilitation (P [is less than] .01). The absolute mean changes (as a percentage of full scale) for the DHI scales and the SF-36 scales are shown in Table 6. The total DHI score changed 11.94%, while the PCS-36 and MCS-36 scores changed 6.67% and 4.35%, respectively.

Table 6. Mean Percentage of Change in Dizziness Handicap Inventory[16] (DHI) and Medical Outcomes Study 36-Item Short-Form Health Survey[27,29] (SF-361 Scores Between Test and Retest(a)

                                 Percentage of
                                    Change
Variable (Possible Range)          X      SD     Minimum   Maximum
DHI
 1. Total (100)                  11.94   15.60   -18.00     54.00
 2. Functional (100)             15.59   19.79   -22.22     66.67
 3. Physical (100)               12.21   18.46   -21.43     57.14
 4. Emotional (100)               8.06   16.34   -22.22     44.44
SF-36
 Physical function (100)(1-3b)   12.74   19.36   -20.00     75.00
 Role limitation (physical)
   (100)(1-3b)                   23.39   48.71   -75.00    100.00
 Bodily pain (100)               10.13   20.46   -22.00     80.00
 General health (100)            -1.90   15.41   -30.00     40.00
 Vitality/energy (100)(4)        10.00   18.79   -20.00     55.00
 Social function (100)(1-4c)     18.14   25.99   -25.00     75.00
 Role limitation (emotional)
   (100)(4)                      16.13   52.26   -66.67    100.00
 Mental health (100)(4*)          0.71   14.13   -32.00     32.00
 Physical component summary
   (74 55)(1-3b*)                 6.67    9.44   -13.26     30.31
 Mental component summary
   (70.73)(1,4d)                  4.35   15.02   -20.84     33.76

(a) Mean change (% of full scale)=absolute change/total possible range x 100. superscript numbers show which SF-36 variables were compared with respective DHI variables. Astersik (*) indicates significant difference with specified variables at specified probability value.

(b) Probability value for multiple comparisons = .0167 = (.05/3).

(c) Probability value for multiple comparisons=.0125= (.05/4).

(d) Probability value for multiple comparisons=.025= (.05/2).

Results of the dependent t tests comparing the changes in DHI scores with changes in SF-36 scores are presented in Table 6. The change in the DHI functional score (15.59%) was greater than the change in the PCS-36 score (6.67%) (P=.006). The change in the mental health score (0.71%), which was lower than the change in the DHI emotional score (8.06%) (P=.016), was the only individual SF-36 score that was different from the DHI scores.

The numbers of patients required to detect the measured change for the DHI and SF-36 are presented in Table 7. The DHI functional scale was the most responsive scale, requiring only 7.24 patients to detect the measured clinical change. Physical function was the most responsive SF-36 scale, requiring 22.07 patients to detect the measured change. The total DHI and PCS-36 required 9.43 and 25.87 patients, respectively, to detect measured changes. Overall, the DHI appeared to be more responsive than the SF-36.

Table 7. Responsiveness (Guyatt's) for the Dizziness Handicap Inventory[16] (DHI) and Medical Outcomes Study 36-Item Short-Form Health Survey[27,29] (SF-36) Scores

                    Mean Change     Variability in
                    (Retest-Test)   Stable Patient(a)
Variable              (n=31)          (n=20)

DHI
 Total                  11.94            7.19
 Functional             15.59            8.23
 Physical               12.21           16.37
 Emotional               8.06            7.07

SF-36
 Physical function      12.74           11.74
 Role limitation
   (physical)           23.39           33.71
 Bodily pain            10.12           12.36
 General health        - 1.90           14.96
 Vitality/energy        10.00           10.19
 Social function        18.14           14.32
 Role limitation
   (emotional)          16.13           24.84
 Mental health           0.71            9.00
 Physical component
   summary               4.97            4.96
 Mental component
   summary               3.08            4.43

                              Guyatt's
                              Responsiveness
Variable                      Statistic[28]     Sample Size(b)

DHI
 Total                           1.66               9.43
 Functional                      1.89               7.24
 Physical                        0.75              46.71
 Emotional                       1.14              19.97

SF-36
 Physical function               1.08              22.07
 Role limitation
   (physical)                    0.69              54.00
 Bodily pain                     0.82              38.70
 General health                  -0.1             316.06
 Vitality/energy                 0.98              26.99
 Social function                 1.27              16.19
 Role limitation
   (emotional)                   0.65              61.64
 Mental health                   0.08              41.80
 Physical component
   summary                       1.00             225.87
 Mental component
   summary                       0.70              53.72

(a) Stable patient variability was calculated from the variability of the patients in the reliability phase of the study.

(b) Sample size=2X[(3.605)X(variability/clinical change)[2]]. Sample size calculated for independent groups ([Alpha] =.05, [Beta] 3=.10).

The SF-36 scores for the patients with vestibular dysfunction before entering the vestibular rehabilitation program and after 6 to 8 weeks of vestibular rehabilitation (n=31) are shown with the SF-36 normative general population data[27] in the Figure (upper and lower panels). The SF-36 scores for the 31 patients prior to vestibular rehabilitation were not different from the SF-36 scores for the larger group of 95 patients seen before vestibular rehabilitation. Because the 31 patients were a subset of the 95 patients evaluated before vestibular rehabilitation, an additional analysis comparing the 31 patients and the other 64 patients indicated that there was no difference in their SF-36 scores. The general population data were taken from the National Survey of Functional Health Status,[27] a mail and telephone survey that include the SF-36. The 2,474 patients in that survey varied in age from 18 to 94 years; 57% were female and 71% were under the age of 65 years.[27] The 31 patients with vestibular disease in the responsiveness component of our study varied in age from 34 to 88 years; 58% were female and 61% were under the age of 65 years. As depicted in the Figure, patients entering a vestibular rehabilitation program had lower scores on the SF-36 scales than did the general population (P [is less than] .05). At follow-up, however, the patients receiving vestibular rehabilitation had scores that were generally closer to those of the normative general population. In particular, role limitation (physical) and social function scales of the SF-36 were greatly improved (P=.012 and P=.0005, respectively).

Discussion

Ceiling and floor effects interfere with a tests ability to provide clinically useful information about function of patients at a single point in time and over time. A test should have little to no ceiling and floor effects to be useful. Increasing the range of health status improves the precision of the instrument.[26] The DHI had minimal ceiling and floor effects, whereas the SF-36 scores showed ceiling or floor effects for some scales (Tab. 1). Floor effects were present for the role limitation (physical) and role limitation (emotional) scales and were greater than the 15% recommended by McHorney and Tarlov[31] for use with individual patients. Ceiling effects were present in the bodily pain scale (18.9%), the social function scale (14.7%), and the role limitation (emotional) scale (25.3%) in this study.

The relationship between the DHI and the SF-36 was similar to other comparisons between disease-specific and generic quality-of-life measures.[20,32] The DHI and the SF-36 were not highly correlated. The highest correlation between the two questionnaires (ie, between the DHI emotional scale and SF-36 social function scale) was .71. Similarities between these two scales are more obvious after reviewing the questions. The DHI emotional scale questions include "Because of your problem, are you afraid to leave home without someone accompanying you?" and "Has your problem placed stress on your relationship with members of your family or friends?" The SF-36 social function scale includes the question "During the past 4 weeks, how much of the time has your physical or mental health interfered with social activities?" Thus, the DHI and the SF-36 may reflect different concepts, which may be complementary.

Our investigation, as well as other studies,[20,32] demonstrated that correlations between disease-specific and generic instruments vary, depending on the instruments. Scales such as the physical function scale may be highly correlated with disease-specific instruments.[20] In our study, the SF-36 physical function scale showed one of the highest correlations to the DHI functional scale (r=.65). Additional studies are needed to validate this finding by comparing both measures with impairment or performance tests.

Another issue that may have affected the relationship between the DHI and the SF-36 is the construct of the DHI. Further analysis of the correlations of the individual DHI scales indicated that the individual DHI scales may be cross correlated and not measure one particular health concept. Fifty-five percent of the variability in the scores for the DHI functional scale was accounted for by the DHI emotional scale. Forty-five percent of the variability in the DHI functional scale was accounted for by the DHI physical scale. Rasch analysis,[33] which was beyond the scope of this study, may be used to determine the unidimensionality and reproducibility of each item's position along a scale. Thus, the scaling properties of the DHI warrant further investigation.

The test-retest reliability of the self-administered DHI was similar to the reliability reported by Jacobson and Newman[16] for their face-to-face interview with 14 patients. Jacobson and Newman reported that same-day test-retest reliability varied from .92 for the DHI physical scale scores to .97 for the DHI emotional and total scale scores. In our study, the DHI showed a similar level of reliability, suggesting that the DHI physical, emotional, and total scores may be important measurements in clinical practice.

Examination of the data showed that the low ICC for the SF-36 role limitation (physical) scale may have been related to how the patients interpreted the single question making up this health concept. Two patients had opposite scores on the test and the retest. The patients may have misread the question or may have changed their response due to education and increased awareness of their conditions after their physical therapy evaluation. In addition, the distribution of the SF-36 role limitation (physical) scores was not normal. Twelve patients had a score of 0. Although the distribution was skewed, Streiner and Norman[34] noted that the ICC statistic is robust even for unequal distributions. The Spearman rank correlation (r=.65) was similar to the Pearson product-moment correlation (r=.64) for the SF-36 role limitation (physical) scale.

The test-retest reliability of the SF-36 for patients with vestibular dysfunction was similar to the test-retest reliability found for other patient populations,[35-37] further verifying the generalized reproducibility of the SF-36. Direct comparisons were difficult because other studies did not provide the range of scores for the test-retest reliability sample and the time interval for administering the test varied. The SEMs calculated in our study also provided additional information regarding the degree of error in the SF-36 and the DHI. Although both the DHI and the SF-36 had reliable measurements, the variability in the SF-36 and DHI scores requires relatively large changes before a therapist could be sure that true change has occurred.

The purpose of testing responsiveness was for us to determine whether the DHI and SF-36 scores changed equally over time. All the DHI variables and most of the SF-36 variables, except for the MCS-36 and the general health, role limitation (emotional), mental health scales, showed changes between the baseline assessment and the assessment done after approximately 6 to 8 weeks of rehabilitation. Because there was no control group, we do not know whether this change was due to physical therapy, medical intervention, or spontaneous remission. Randomized controlled clinical trials are under way in an attempt to understand the effect of specific vestibular rehabilitation protocols on this change in health status. For our study, vestibular rehabilitation was not precisely defined. In general, the treatments were home programs consisting of exercises to assist with balance and reduce motion sensitivity. There is a wide array of treatments routinely recommended for each of the diagnostic groups comprising vestibular pathology.

Two factors may have affected the use of the dependent t test in determining whether the two questionnaires were equally responsive. First, an assumption was made that each questionnaire had similar scaling properties when the absolute raw score taken as a percentage of full scale was used for analysis. For example, a 10-point change in scores on the SF-36 was equivalent to a 10-point change in scores on the DHI, given that the scores for each scale ranged from 0 to 100. Further studies are needed to determine whether this scaling assumption is true. The method of adjusting the PCS-36 and MCS-36 scores, however, appears to be more valid than taking a true percentage change (pretest score - posttest score/pretest score x 100) where an identical absolute change in scores is highly influenced by the baseline (pretest) score. Second, the total score range for the PCS-36 and MCS-36 is not from 0 to 100. The true range was estimated by entering the highest possible and lowest possible SF-36 scores into the formulas used to calculate the component summary scores. This normalization was needed so that absolute changes in the PCS-36 and MCS-36 scores could be compared with the changes in scores for the DHI (0-100). An alternate method would be to use the range of scores for the patients in our study instead of using the highest and lowest possible scores. A separate analysis using the range of scores from our study showed no differences in changes in scores between the DHI and the SF-36; thus, the SF-36 and DHI would be considered equally responsive. We contend, therefore, that the method of defining the range for the PCS-36 and MCS-36 scores contributed to the findings of our study.

Responsiveness has not been previously determined for the DHI and the SF-36 in patients with vestibular dysfunction. Comparisons between generic and disease-specific instruments, however, have shown similar results. In a study comparing the Roland Disability Questionnaire with the SF-36, Guyatt's responsiveness statistic was compared in 318 patients who had an improvement in their sciatic pain symptoms over 3 months.[21] The SF-36 physical function scale (2.3) and bodily pain scale (2.0) were more responsive than the Sciatic Frequency Index (- 1.6). The SF-36 physical function scale (2.3) was more responsive than the modified Roland disease-specific scale (- 2.0). The responsiveness was negative because a decrease in scores represented an improvement in symptoms. Like our vestibular study, this study showed that some of the SF-36 scales were more responsive than the disease-specific scales.

An interesting finding was the similarity in responsiveness between the SF-36 component summary scores and the SF-36 individual scores. The responsiveness of the PCS-36 scores (1.00) was similar to that of the SF-36 physical function scores (1.08). The use of the PCS-36 is supported by this finding because its sensitivity to change was good, even though it factors in several aspects of physical health. Thus, a global representation of the physical health domain may be obtained without losing test score sensitivity. McHorney and Tarlov[31] suggested that an additional advantage of the PCS-36 was standard deviations of the scores were smaller, resulting in a better likelihood that the scale may be used with individuals. A disadvantage, however, is that clinically relevant information about the patients' function and well being may be lost.[31] All of these issues should be considered when using the component score to monitor clinical outcomes.

A minimal change should be used for the numerator of the Guyatt statistic when calculating responsiveness.[28] The time internal for change used for Guyatt's responsiveness statistic was an estimate based on previous clinical studies[13,14] and may vary based on the severity of an individual's dysfunction and diagnosis. Thus, the responsiveness in our study is only a preliminary indication of responsiveness of the DHI and the SF-36 for patients receiving vestibular rehabilitation.

Patients with vestibular dysfunction who are referred for vestibular rehabilitation have lower SF-36 scores than does the general population.[27] The population of individuals with vestibular disease represents many diagnoses, so these measurements were restricted to the case mix making up our study (Tab. 1). Analysis of the data of the 31 patients evaluated before and after 6 to 8 weeks of rehabilitation showed that their SF-36 scores were similar to those of the other 64 patients, who were evaluated only once. Six to 8 weeks after the baseline evaluation, however, the scores of the 31 patients were improved but remained lower than the general population means (Figure). Direct comparison with general population data is limited because the age range varies and other confounding variables such as weight, height, other comorbid conditions could not be compared. The length of illness in the patients in our study ranged from 1 to 120 months, with a mean of about 26 months. Although this group represents a broad spectrum of individuals with vestibular dysfunction, these SF-36 scores suggest that individuals referred for vestibular rehabilitation have an extremely poor perception of their health-related quality of life.

[Figure ILLUSTRATION OMITTED]

Because normative descriptive data have not been developed for the DHI, comparisons between our subjects and the general population are not possible. The average DHI scores for the patients in our study, however, were lower than those reported for 101 patients in another study.[38] Differences in scores between the two studies may be attributed to the stage of disease or different diagnostic groups of patients.[38] For example, 66% of the patients in Robertson and Ireland's study[38] had peripheral vestibular dysfunction, whereas 86% of the patients in our study had peripheral vestibular dysfunction. Thus, the case mix must be considered when interpreting the extent of the perceived disability.

Summary and Conclusions

Health-related quality-of-life questionnaires, in addition to impairment and human performance tests, may provide physical therapists with information to determine true treatment efficacy. Both the DHI and the SF-36 demonstrated good between-day reliability in patients with vestibular dysfunction. Based on relatively low correlations, the DHI and the SF-36 appear to provide different but complementary information about the health status of patients with vestibular dysfunction. Overall, the DHI total score was more responsive than the PCS-36 and the MCS-36. As in previous studies, however, the disease-specific instrument was not always more responsive than the generic instrument in all domains. Relative to a normative database, patients referred for vestibular rehabilitation have a poor perception of their health status. These health status assessments were highly reproducible 24 to 48 hours apart, and the patients' health status was improved after 6 to 8 weeks of vestibular rehabilitation. Whether vestibular rehabilitation contributed to that improvement warrants further study.

Acknowledgments

We thank Dr Jay Rubinstein for his assistance with confirming the patients' vestibular classification. We thank Mary Lohse Shepherd, PT, and the Department of Otolaryngology, University of Iowa Hospitals and Clinics, Iowa City, Iowa, for subject recruitment and questionnaire administration. We acknowledge Carol Leigh for her assistance with the preparation of the manuscript.

(*) SAS Institute Inc, PO Box 8000, Cary, NC 27511-8000.

References

[1] Sloane P, Blazer D, George L. Dizziness in a community elderly population. J Am Geriatr Soc. 1989;37:101-108.

[2] Nedzelski J, Barber H, McIlmoy I., Diagnosis in a dizziness unit. J Otolaryngol. 1986;15:101-104.

[3] Kroenke K, Lucas (2, Rosenberg M, Schrokman B. Psychiatric disorders and functional impairment in patients with persistent dizziness. J Gen Intern Med. 1993;8:530-535.

[4] McGee S. Dizzy patients: diagnosis and treatment. West J Med. 1995;162:37-42.

[5] Baloh RW. Dizziness, Hearing Loss, and Tinnitus: The Essentials of Neurotology. Philadelphia, Pa: FA Davis Co; 1984:159-187.

[6] Yardley L, Masson E, Verschuur C, et al. Symptoms, anxiety, and handicap in dizzy patients: development of the vertigo symptom scale. J Psychom Res. 1992;36:731-741.

[7] Yardley L, Putman J. Quantitative analysis of factors contributing to handicap and distress in vertiginous patients: a questionnaire study. Clin Otolaryngol. 1992;17:231-236.

[8] Yardley L, Todd AM, Harter-Lacoudraye MM, Ingham R. Psychosocial consequences of recurrent vertigo. Psychology and Health. 1992;6: 85-96.

[9] Boult C, Murphy J Sloane P, et al. The relation of dizziness to functional decline. J Am Geriatr Soc. 1991;39:858-S61.

[10] Cohen H, Kane-Wineland M, Miller L., Hatfield C. Occupation and visual/vestibular interaction in vestibular rehabilitation. Otolaryngol Head Neck Surg. 1995;112:526-532.

[11] Norre ME, Beckers A. Vestibular habituation training: exercise treatment for vertigo based upon the habituation effect. Otolaryngol Head Neck Surg 1989;101:14-19.

[12] Shumway-Cook A, Horak FB. Vestibular rehabilitation: an exercise approach to managing symptoms of vestibular dysfunction. Seminars in Hearing. 1989;10:196-208.

[13] Horak FB, Jones-Rycewicz C, Black FO, Shumway-Cook A. Effects of vestibular rehabilitation on dizziness and imbalance. Otolaryngol Head Neck Surg 1992;106:175-180.

[14] Krebs DE, Gill-Body KM, O'Riley P, Parker SW. Double-blind, placebo controlled trial of rehabilitation for bilateral vestibular hypofunction: preliminary report. Otolarngol Head Neck Surg 1993;109:735-741.

[15] Mruzek M, Marin K, Nichols D, et al. Effects of vestibular rehabilitation and social reinforcement on recovery following ablative vestibular surgery. Laryngoscope. 1995;105:686-692.

[16] Jacobson GP, Newman CW. The development of the dizziness handicap inventory. Arch Otolarngol Head Neck. Surg. 1990;116:424-427.

[17] Patrick DL, Deyo RA. Generic and disease-specific measures in assessing health status and quality of life. Med Care. 1989;27:S217-S232.

[18] Ware JE Jr. Methodological considerations in the selection of health status assessment procedures. In: Wenger N, Mattson M, Furberg C, Elinson J, eds. Assessment of Quality of Life in Clinical Trials of Cardiovascular Therapies. New York, NY: Le Jacq Publishing Inc; 1984:87-111.

[19] Guyatt G, Feeny D, Patrick DL. Issues of quality of life measurement in clinical trials. Control Clin Trials. 1991;12:81S-90S.

[20] Kantz ME, Harris W, Levitsky K, et al. Methods for assessing condition-specific and generic functional status outcomes after total knee replacement. Med Care. 1992;30:MS240-MS252.

[21] Patrick DL, Deyo RA, Atlas SJ, et al. Assessing health-related quality of life in patients with sciatica. Spine. 1995;20:1899-1909.

[22] Ware JE Jr, Sherbourne CD. The MOS 36-Item Short-Form Health Survey (SF-36), I: conceptual framework and item selection. Med Care. 1992;30:473-481.

[23] Deyo RA, Diehr P, Patrick DL. Reproducibility and responsiveness of health status measures: statistics and strategies for evaluation. Control Clin Trials. 1991;12:142S-158S.

[24] Jette AM. Using health-related quality-of-life measures in physical therapy outcomes research. Phys Ther. 1993;73:528-537.

[25] Wagner E, LaCroix A, Grothaus L, Hecht J. Responsiveness of health status measures to change among older adults. J Am Geriatr Soc. 1993;41:241-248.

[26] McHorney CA, Ware JE Jr, Rachel Lu JF, Sherbourne CD. The MOS 36-item Short-Form Health Survey (SF-36), III: tests of data quality, scaling assumptions, and reliability across diverse patient groups. Med Care. 1994;32:40-66.

[27] Ware JE Jr, Snow K, Kosinski M, Gandek B. SF-36 Health Survey: Manual and Interpretation Guide. Boston, Mass: The Health Institute, New England Medical Center; 1993.

[28] Guyatt G, Walter S, Norman GO. Measuring change over time: assessing usefulness of evaluative instruments. J Chronic Dis. 1987;40: 171-178.

[29] Ware JE Jr, Kosinski M, Keller SD. SF-36 Physical and Mental Health Summary Scales: A User's Manual. Boston, Mass: The Health Institute, New England Medical Center; 1994.

[30] Newman CW, Jacobson GP. Application of self-report scales in balance function handicap assessment and management. Seminars in Hearing 1993;14:363-376.

[31] McHorney CA, Tarlov AR. Individual-patient monitoring in clinical practice: Are available health status surveys adequate? Qual Life Res. 1995;4:293-307.

[32] Bombadier C, Melfi C, Paul J, et al. Comparison of a generic and a disease-specific measure of pain and physical function after knee replacement surgery. Med Care. 1995;33:AS131-AS144.

[33] Haley SM, McHorney CA, Ware JE Jr. Evaluation of the MOS SF-36 physical functioning (PF-10), 1: unidimensionality and reproducibility of the Rasch item scale. J Clin Epidemiol. 1994;47:671-684.

[34] Streiner DL, Norman GO. Health Measurement Scales: A Practical Guide to Their Development and Use. Oxford, England: Oxford University Press; 1995:104-126.

[35] Brazier JE, Harper R Jones NMB, et al. Validating the SF-36 health survey questionnaire: new outcome measure for primary care. BMJ 1992;305:160-164.

[36] Jette DL, Downing J. Health status of individuals entering a cardiac rehabilitation program as measured by the Medical Outcomes Study 36 Item Short-Form Survey (SF-36). Phys Ther. 1994;74:521-527.

[37] Vickrey BG, Hays RD, Harooni R, et al. A health-related quality-of-life measure for multiple sclerosis. Qual Life Res. 1995;4:187-206.

[38] Robertson D, Ireland D. Dizziness handicap inventory correlates of computerized dynamic posturography. J Otolaryngol. 1995;24:118-124.

Appendix 1. Dizziness Handicap Inventory[16]

Patient Instructions: The purpose of this scale is to identify difficulties that you may be experiencing because of your dizziness or unsteadiness. Please answer "yes," "no," or sometimes to each question.

Functional Scale

Because of your problem do you have difficulty
getting into or out of bed?

Does your problem restrict your participation
in social activities such  as going out to
dinner going to the movies dancing or
going to parties?

Because of your problem do you have
difficulty reading?

Because of your problem do you avoid heights?

Because of your problem is it difficult for you
to do strenuous housework or yardwork?

Because of your problem is it difficult for you
to walk around your house
in the dark?

Does your problem interfere with your job or
household responsibilities?

Because of your problem do you restrict your
travel for business or recreation?

Physical Scale

Does looking up increase your problem?

Do quick movements of your head increase
your problem?

Does turning over in bed increase your
problem?

Does walking down a sidewalk increase
your problem?

Does bending over increase your problem?

Does performing more ambitious activities
such as sports dancing or household
chores (eg sweeping or putting away
dishes) increase your problem?

Emotional Scale

Because of your problem do you feel
frustrated?

Because of your problem are you afraid to
leave your home without having someone
accompany you?

Because of your problem have you been
embarrassed in front of others?

Because of your problem are you afraid that
people may think that you are intoxicated?

Because of your problem is it difficult for you
to concentrate?

Because of your problem are you afraid to
stay home alone?

Because of your problem do you feel
handicapped?

Has your problem placed stress on your
relationship with members of your family or
friends?

Because of your problem are you
depressed?

Appendix 2
Summary of Medical Outcome Study 36-item Short-Form Health
Survey[27,29] (SF-36) Scales

Scale                         Low Score Meaning

Physical function             Limited a lot due to health
                              in performing all physical
                              activities including walking
                              dressing and bathing

Role limitation due           Problems with work or other
to physical problems          daily activities due
                              to physical health

Bodily pain                   Very severe and limiting pain

General health                Evaluates personal health as
                              poor and believes
                              that it is likely to get worse

Social function               Extreme and frequent
                              interference with social
                              activities due to emotional
                              or physical problems

Vitality/energy               Feels tired and worn out
                              all the time

Role limitation due to        Problems with work or
emotional problems            other activities due to
                              emotional problems

Mental health                 Feelings of nervousness
                              and depression all of the
                              time

Physical component summary    Substantial limitations in
                              self-care and physical social
                              and role activities: severe
                              bodily pain frequent
                              tiredness health rated as poor

Mental component summary      Frequent psychological distress
                              substantial social and role
                              disability due to emotional
                              problems; health in general
                              rated as poor

Scale                         High Score Meaning

Physical function             Ability to perform all
                              types of activities
                              including the most
                              vigorous activities
                              without limitations due
                              to health

Role limitation due           No problems with work or
to physical problems          other daily activities as a
                              result of physical health

Bodily pain                   No pain or limitations
                              due to pain

General health                Evaluates personal health
                              as excellent

Social function               Performs normal social
                              activities without
                              interference due to
                              physical or emotional
                              problems

Vitality/energy               Feels full of energy all
                              the time

Role limitation due to        No problems with work
emotional problems            or other daily
                              activities as a result
                              of emotional health

Mental health                 Feels peaceful calm and
                              happy all the time

Physical component summary    No physical limitations
                              disabilities or
                              decrements in well being;
                              high energy level; health
                              rated as excellent

Mental component summary      Frequent positive affect;
                              absence of psychological
                              distress and limitation in
                              usual social or role
                              activities due to
                              emotional problems; health
                              rated as excellent

LJ Enloe, PT, is Physical Therapist, Veterans Administration Medical Center, Iowa City, Iowa. She was a student in the Physical Therapy Graduate Program, The University of Iowa, when this research was completed in partial fulfillment of the requirements for her Master of Arts degree in physical therapy.

RK Shields, PhD, PT, is Assistant Professor, Physical Therapy Graduate Program, College of Medicine, The University of Iowa, 2600 Steindler Bldg, Iowa City, IA 52242-1008 (USA) ([e-mail]), and Clinical Research Coordinator, University of Iowa Hospitals and Clinics, Iowa City, Iowa. Address all correspondence to Dr Shields at the first address.

This study was approved by The University of Iowa College of Medicine Human Subjects Review Committee.

This study was supported in part by The University of Iowa Physical Therapy Clinical Research Center, which was originally funded by the Foundation for Physical Therapy Inc.