The Tinnitus Functional Index

2012, Ear and Hearing

visibility

…

description

24 pages

link

1 file

Abstract

Chronic subjective tinnitus is a prevalent condition that causes significant distress to millions of Americans. Effective tinnitus treatments are urgently needed, but evaluating them is hampered by the lack of standardized measures that are validated for both intake assessment and evaluation of treatment outcomes. This work was designed to develop a new self-report questionnaire, the Tinnitus Functional Index (TFI), that would have documented validity both for scaling the severity and negative impact of tinnitus for use in intake assessment and for measuring treatment-related changes in tinnitus (responsiveness) and that would provide comprehensive coverage of multiple tinnitus severity domains.

Figures (13)

*To our knowledge, information on patient base has not been published. TABLE 1. Nine widely used tinnitus questionnaires and the corresponding clinical samples on which they were based

TABLE 2. Important design considerations in constructing the Tinnitus Functional Inde

*Site did not participate in collection of Prototype 1 follow-up data. tSite did not participate in collection of any Prototype 2 data. TFI, Tinnitus Functional Index. TABLE 3. Number of participants providing initial and follow-up data (3 mo, 6 mo) at each site for TFI Prototypes 1 and 2

TABLE 4. Sample characteristics for Tinnitus Functional Index Prototypes 1 and 2 at initial and follow-up data collection

TABLE 4. Continued. For any characteristic, when numbers for the initial sample do not total to 326 (Prototype 1) or 347 (Prototype 2) or percentages do not total 100%, it is because of missing data (e.g., blank, refused, don’t know). *Minority or mixed race for Prototype 1 included 22 individuals, with specific race categories indicated (participants could check more than one): African American or Black (13), American Indian/Alaska Native (9), Asian (2), Hawaiian or Pacific Islander (1), White (8), Other (1). Minority or mixed race for Prototype 2 included 31 individuals, with specific race categories indicated (participants could check more than one): African American or Black (16), American Indian/Alaska Native (5), Asian (2), Hawaiian or Pacific Islander (1), White (4), Other (7). tFor Prototype 1, “Explanation and information about tinnitus” was not specified as a treatment option in the questionnaire but was added for Prototype 2.

TABLE 5. Comparison of performance of three versions of the TFI Mean (SD) values are rounded to whole numbers. BDI-PC, Beck Depression Inventory-Primary Care; TFl, Tinnitus Functional Index; THI, Tinnitus Handicap Inventory; VAS, Visual Analog Scale. *For Prototype 1, factor analysis excluded participants who reported “not at all” to how much of a problem their tinnitus is. For Prototype 2 and final TFl, factor analysis excluded participants who reported “not at all” or “small” to the same question. tFor Prototype 1, 6 mo effect sizes were not computed because of small group sizes.

Pooled SD combines SDs for initial and 3 mo scores. In preliminary presentations of this work (Meikle, Henry et al. 2008; Meikle, Stewart et al. 2008), computations of effect sizes took advantage of the repeated-measures design by computing effect sizes using initial minus follow-up mean differences, divided by the SD of the difference scores. Later, we adopted a different method for calculating effect sizes, changing the denominator to the pooled SD of the initial and follow-up scores. The latter method makes the effect sizes of repeated-measures designs comparable to the effect sizes obtained in studies comparing different groups of subjects (e.g., comparing treated with untreated subjects), thus facilitating meta-analyses that may be conducted in the future. TF, Tinnitus Functional Index; THI, Tinnitus Handicap Inventory; VAS, Visual Analog Scale. *Not all of the 65 follow-up participants completed all three of the questionnaires: 64 participants completed TFI Prototype 1, 64 completed the THI, and 57 completed the VAS. TABLE 6. Comparison of mean change at 3 mo follow-up for improved, unchanged, or worse groups on overall scores for TFI Prototype 1 (30 of 43 items retained for Prototype 2), Tinnitus Handicap Inventory, and severity VAS

The analysis was computed using 283 Prototype 2 participants who reported that their tinnitus problem was a moderate problem, a big problem, or a very big problem. There were 11 additional participants who reported that their tinnitus problem was a moderate, big, or very big problem, but were excluded from the analysis because they were missing responses to more than 10% (three or more items) of the final 25-item TFI. The factor analysis was then computed using mean substitution of missing data for the 283 participants. Bold values indicate factor loadings whose absolute values are 0.30 or greater. Table entries are factor loadings from the pattern matrix and are standardized regression coefficients (beta weights) for predicting each item from the eight factors. The percentage of variance explained by the eight factors was 79.5%. TFI, Tinnitus Functional Index; h?, communality (percentage of variance in each item explained by the eight factors). TABLE 7. Results of an eight-factor solution using principal-axis factor analysis with oblique rotation of 25 items from Prototype 2 that were retained for the final TFI

TABLE 8. Effect size estimates for the overall scales (final 25-item TFI, VAS, THI) and subscales (TFI, THI) at 3 and 6 mo follow-ur Perceived Change in Overall Tinnitus Condition From Initial Visit to 6 mo Follow-Up

Regarding missing data at 3 mo, 5 of the 155 participants providing 3 mo follow-up data did not answer the perceived-change item, so the total sample size used for this table was 150. In estimating effect sizes for each of the five perceived-change levels, no scale or subscale (except for the VAS) had more than one participant missing a score. For the VAS, 22 participants had a missing score, with the number missing in each perceived-change level as follows: 1 in much or moderately improved, 4 in slightly improved, 13 in unchanged, 0 in slightly worse, and 4 in moderately or much worse. Regarding missing data at 6 mo, 1 of the 86 participants providing 6-mo follow-up data did not answer the perceived-change item, so the total sample size used for this table was 85. As with 3 mo, in estimating effect sizes for each of the five perceived change levels, no scale or subscale (except for the VAS) had more than one participant missing a score. TFI, Tinnitus Functional Index; THI, Tinnitus Handicap Inventory; VAS, Visual Analog Scale.

1988). These estimates agree well with a recent meta- analytic review by Norman et al., in which they concluded “under many circumstances, when patients with a chronic disease are asked to identify minimal change, the estimates fall very close to half a SD” (Norman et al. 2003, p. 590). Pending further clinical observations using the TFI, which may help to refine our estimate, we suggest that a reduction of 13 points be used as the criterion for meaningful reduction in TFI outcome scores. the large effect sizes for the THI Catastrophic subscale. An important difference between the TFI and THI is that the THI’s domain of catastrophic responses was not included in the TFI. The Tinnitus Research Consortium, which funded this study, had requested that the TFI minimize negative ideation and exclude catastrophic items. One reason for omitting such items is that persons with mild tinnitus sometimes become concerned that such catastrophic feelings will be their fate eventually, creating a negative starting point for intervention. The five items on the THI Cata- strophic subscale focus on patients’ feelings of desperation, inability to escape from tinnitus, fear of having a terrible disease, loss of control, and inability to cope. Two of the three items on the new TFI Sense of Control subscale have

Fig. 3. Frequency distributions of overall TFI scores after stratification by responses to the item “How much of a problem is your tinnitus?” Horizontal axis shows overall TFI scores (in 10 bins ranging from 0 to 100 for each of the five small graphs) for each of five groups of participants identified by their responses fo the tinnitus problem question, ranging from those responding “not a problem” to those responding “a very big problem.” Vertical axis shows the percentage of participants’ responses occurring in each bin of the five successive frequency distributions. Numbers at upper right corner in each graph display the group sizes together with the group mean TFI scores and SDs. TFI, Tinnitus Functional Index. Use of TFI and THI Subscales ¢ The effect-size estimates in Table 8 show that subscales differed considerably from one another, with some subscales exhibiting effect sizes that were two to three times larger than those of other subscales. An unexpected finding regarding subscale effect sizes was

Fig. 4. Overall mean TFI change scores at 3 and 6 mo follow-up after stratification by participants’ Global Perception of Change. Horizontal axis shows five levels of participants’ Global Perception of Change. Vertical axis shows mean change in scores for overall TFI (follow-up score minus initial score) for each of the two follow-up intervals (3 and 6 mo). Using an analysis of variance to compare group means, both the 3 and 6 mo means displayed significant overall differences, resulting in F ratios of 10.23 at 3 mo and 12.96 at 6 mo (p < 0.001 for both). TFI, Tinnitus Functional Index.