Professional Documents
Culture Documents
Dhi Penting Daftar Pustaka
Dhi Penting Daftar Pustaka
Objective.: The Dizziness Handicap Inventory (DHI) is a with three grouping factors (i.e., Physical manifestations,
25-item self-report questionnaire developed to measure the Catastrophic impact of dizziness, and the Emotional impact
disabling and handicapping impact of dizziness. The present of dizziness) were fit to a different random sample of 992
investigation was conducted in an effort to re-assess the patients using the new item-to-group factor specifications.
factor structure of the DHI. Results.: In the confirmatory analyses, all items had a
Study Design.: Retrospective study. positive factor loading on the general factor. There were 14
Setting.: Tertiary care center. items that loaded on the general factor only. The rest of the
Patients.: Subjects were 1,991 patients who were evaluated items (n ¼ 11) loaded on both the general factor and one of
in the Mayo Clinic-Rochester Vestibular and Balance three group factors.
Laboratory. Conclusions.: Conclusions of the study revealed several
Main Outcome Measures.: Exploratory factor analysis: an findings: 1) reporting the result as a total score (i.e., a single
exploratory bifactor analysis (EFA) with bifactor rotation general factor) is warranted, and, 2) there is statistical
was used to analyze a random sample of 999 patients. support for the existence of three subscales representing: the
Analyses were used to determine the dominance of the Physical manifestations, Catastrophic impact, and Emotional
general factor (i.e., total score) relative to the group factor impact of dizziness and vertigo. Key Words: Dizziness—
(i.e., subscales). Confirmatory factor analysis: a confirmatory Dizziness Handicap Inventory—Self-report dizziness
bifactor graded response model was fit with appropriate handicap.
item-to-group relationships that was discovered by our
exploratory analyses. To validate the bifactor model that was
identified with the exploratory analyses, a bifactor model Otol Neurotol 40:1217–1223, 2019.
The presence and magnitude of peripheral and central most widely-used self-report measure of dizziness dis-
vestibular system impairments may be measured using ability/handicap is the Dizziness Handicap Inventory (1).
several conventional vestibular system electroneurodiag- The DHI was originally developed to evaluate the
nostic tests (e.g., videonystagmography and rotary chair precipitating physical factors associated with dizziness
testing). Despite their usefulness for the diagnosis and as well as the functional and emotional consequences of
management of the dizzy patient these tests are insuffi- vestibular impairment. Since its publication in 1990, the
cient for describing the impact a patient’s dizziness has DHI has been referenced more than 1500 times in the
on their well-being and quality of life. In an attempt to scientific literature and has been translated into more than
understand better the functional impact of disorders 17 languages (2–18). Due to its strong validity and test/
causing dizziness in patients, several self-report ques- retest reliability (1) the DHI has become the ‘‘gold stan-
tionnaires addressing dizziness disability/handicap have dard’’ against which similar scales have been compared.
been developed over the last three decades. Perhaps the An a priori assumption by Jacobson and Newman (1)
was that items in the DHI could be subgrouped into
content domains (i.e., subscales). To this end, the inves-
tigators created subscales that were designed to assess the
Address correspondence and reprint requests to Devin L. McCaslin, Physical manifestations (nine items) of dizziness, ver-
Ph.D., Department of Otolaryngology, Mayo Clinic, Gonda 12-400 tigo, and unsteadiness as well as the Functional (seven
AUD, 200 1st Street SW, Rochester, MN; E-mail: mccaslin.devin@
mayo.edu
items) and Emotional (9 items) consequences of vestib-
The authors disclose no conflicts of interest. ular impairment. The investigators did not evaluate the
DOI: 10.1097/MAO.0000000000002365 validity of the empirically-created subscale structure.
Copyright © 2019 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
1218 K. M. VAN DE WYNGAERDE ET AL.
Copyright © 2019 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
THE COMPONENT STRUCTURE OF THE DIZZINESS HANDICAP INVENTORY 1219
that was identified with the exploratory analyses, a bifactor items were associated with difficulty or fear of leaving
model with three grouping factors (i.e., Physical manifestations, home, going for a walk, staying home by oneself.
Catastrophic impact of dizziness, and the Emotional impact of The third group factor explained 6.4% of the vari-
dizziness) were fit to a different random sample of 992 patients ance and consisted of three items that assessed the
using the new item-to-group factor specifications.
To this end, for Investigation 2, subjects were 992 patients
Emotional impact associated with a vestibular
(mean age 56 yrs, sd 18.05 yrs, 599 men) who were evaluated impairment. They are the items with factor loadings
between November 21, 2016 and August 29, 2017 in our found on Table 2 in the ‘‘Group factor 3’’ column.
Vestibular and Balance Laboratory at Mayo Clinic in Roches- The content of these items focused on the effect that
ter, MN. An excellent model fit and large item-to-group factor dizziness-related frustration, depression, and stress
loadings would indicate that the current bifactor model would that affected interpersonal relationships.
be favored over the original three factor models. Factor 4 explained 4.7% of the variance and consisted
of two items, the content of which centered on activity
RESULTS restrictions. These items may be found in Table 2 in the
‘‘Group factor 4’’ column. These items included the
Exploratory Factor Analysis words ‘‘restrict,’’ which appeared to be a linguistic
Before fitting the exploratory bifactor model, we feature rather than a specific subdomain of dizziness.
searched for evidence of sufficient unidimensionality. And because a factor with only two items is generally
Assuming three factors (i.e., physical, emotional, and func- considered weak or unstable, we limited our group
tional) that were originally proposed for scoring, we found factors to the first three individual factors.
the ECV value was 0.80. Note that ECV is the ratio of
variance explained by the general factor divided by the Confirmatory Factor Analysis
variance explained by the general plus the group factors. In the confirmatory analyses, all items had a positive
With the same assumption, we also found that the vH value factor loading on the general factor. There were 14 items
was 0.92. This evidence supported the essential unidimen- that loaded on the general factor only (see Table 3). The
sionality necessary for fitting the bifactor models. rest of the items (n ¼ 11) loaded on both the general
To decide the number of factors to extract for explor- factor and one of three group factors. The fit of the new
atory bifactor analyses, we searched for the eigenvalues model was excellent (M2 [239] ¼ 1395.09, p < 0.001,
exceeding the criterion of 1.0. There were five factors root mean square error of approximation ¼ 0.070 [90%
with eigenvalues greater than 1.0. Therefore, we con- confidence interval, 0.066–0.073], Tucker Lewis
ducted EFA extracting one general factor and four group Index ¼ 0.96, comparative fit index ¼ 0.96). In the cur-
factors. These factor loadings from the resulting model rent model, all general factor loadings were greater than
are shown in Table 2. 0.30 except item P13 that demonstrated a general factor
The first factor that we are referring to as the General loading of 0.27. The fact that all group factor loadings
factor, accounted for 70% of the total variance. All item were at least 0.25 enabled a clear interpretation of the
loadings on the general factor exceeded 0.30 except for internal structure of the dizziness construct. Items in each
one item, ‘‘Does turning over in bed increase your of the grouping factors had a positive factor loading for
problem?’’ that had a marginal loading of 0.13. Accord- their respective group only. That is, there were no cross-
ingly, one possible solution was to delete item 13 from loadings for any of the items. These results are shown in
the questionnaire completely. However, because this Table 3.
item had a strong factor loading of 0.62 on the first Factor communality estimates (i.e., the proportion of
group factor, the item was retained for further analysis. the variance in an item explained by the model) ranged
Next, we investigated whether generating subscales from 0.32 to 0.89. The percentage of variance in the data
from the residual data (after extracting the general factor) explained by the general factor was 46%. The proportion
was warranted (i.e., based on the content of the items and of variance explained by the physical manifestations factor
salient factor loadings >0.30 and the content). The first was 6%. The amount explained by the Emotional impact
group factor explained 12.9% of the variance and con- factor of was 3%. The percentage of variance explained
sisted of five items that assessed the Physical manifes- by the Catastrophic impact of vertigo was 2%. The
tations of vestibular impairment. These items and their dominance of the general factor persisted with vH ¼ 0.93.
factor loadings are shown in Table 2 in the ‘‘Group factor In this EFA we identified three group factors. We
1’’ column. The content of these items was related to investigated whether the three group factors added value
whether head movements such as looking up, bending beyond the total score (i.e., whether it was worthwhile to
over, turning over in bed, or getting in and out of bed report subscale scores for the three content areas in
increased the patient’s dizziness. addition to the total score). Subscale scores can be of
The second group factor explained 5.9% of the vari- interest to clinicians due to their potential benefits in
ance and consisted of three items that assessed the providing more detailed information about their patients.
patient’s fear of Catastrophic impact (i.e., a catastrophic A general dizziness score provides an overall assessment
event) associated with their vestibular impairment. These of dizziness, while subscale scores have the potential to
items with their factor loadings can be found on Table 2 allow clinicians to monitor and address specific aspects
in the ‘‘Group factor 2’’ column. The content of these of dizziness. Thus, it can be determined whether subscale
Copyright © 2019 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
1220 K. M. VAN DE WYNGAERDE ET AL.
TABLE 2. Exploratory bifactor analyses and their factor loadings on the general factor and on each of the four subscales
Factor 1 Factor 2 Factor 3 Factor 4
General Physical Emotional Catastrophic Participation
Factor Manifestations Impact Impact Restriction
Only non-trivial group factor loadings (>0.30) are printed. The group factor loadings 0.30 are not presented in this table.
scores from the group factors add value over the total score dizziness, catastrophic impacts of vertigo, and emotional
alone (i.e., general dizziness score) using the concept, effect of vertigo and dizziness subscales.
‘‘proportional reduction in mean squared error’’ (PRMSE).
Here the true subscale score is estimated based on the DISCUSSION
observed subscale score and the observed total score. If
the PRMSE of the subscale score (PRMSEs), which is equal Since the publication of the DHI in 1990, results of
to the reliability of the subscale score is greater than the several investigations employing either FA or PCA have
PRMSE of the total score (PRMSEx), then there is a positive supported the contention that related items in the DHI can
added value of reporting the subscale score. Using this be grouped statistically to provide additional information
method, it appeared that all three subscales contributed beyond the general factor (i.e., effect of dizziness on the
added value. Specifically, the PRMSEs values were 0.75, emotional impact of dizziness on the patient). Results of
0.75, and 0.71 while the corresponding PRMSEx values these investigations generally have supported a two or
were 0.50, 0.69, 0.63 for physical manifestations of three factor solution.
Copyright © 2019 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
THE COMPONENT STRUCTURE OF THE DIZZINESS HANDICAP INVENTORY 1221
TABLE 3. Factor loadings from the confirmatory bifactor model with three group factors (n ¼ 992)
Factor
General Physical Emotional Catastrophic Communality
Factor Manifestations Impact Impact Estimatesa
For example, Asmundson et al. (19) performed a PCA postural instability in relation to contextual factors such
on 95 DHI questionnaires obtained in their vestibular as integrating visual stimuli. This factor was maintained
disorders clinic. The results of their analysis supported in the Dutch version of the DHI (7) and was similar in the
either a 2-factor solution consisting of ‘‘general functional German (20) and Japanese (13) versions. The Japanese
limitations’’ and ‘‘postural difficulties’’ or a 3-factor version also contained four other subscales, two of which
solution of ‘‘disability in activities of daily living,’’ related to the vestibular disability and handicap subscales
‘‘phobic avoidance,’’ and ‘‘postural difficulties.’’ suggested by Perez et al. (16).
Perez et al. (16) conducted a PCA on responses Kurre et al. (20) conducted an investigation of the
obtained from 337 subjects using the Spanish version German translation of the DHI. The investigators evalu-
of the DHI. They identified a 3-factor solution consisting ated 94 subjects. Results of their efforts supported a
of ‘‘vestibular handicap,’’ ‘‘vestibular disability,’’ and 3-factor solution ‘‘effect of dizziness and unsteadiness
‘‘visuo-vestibular disability.’’ The third factor, ‘‘visuo- on emotion and participation,’’ ‘‘specific activities/
vestibular disability,’’ described the self-perceived movements provoking dizziness and unsteadiness,’’
Copyright © 2019 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
1222 K. M. VAN DE WYNGAERDE ET AL.
and ‘‘self-perceived walking ability/feeling of postural The results of the two investigations reported herein
stability in relation to contextual factors.’’ Emulating the have suggested that: 1) if desired, the existing 25-item
data extracted in the Spanish, Dutch, and English ver- total score can remain, ‘‘as is.’’ Additionally, the results
sions, these subscales were comparable in their content suggest that it is possible to create three subscales, from
domains of ‘‘participation restrictions,’’ ‘‘limitations the original 25-item device that address the effects of:
in activities due to motion sensitivity,’’ and ‘‘visuo- changing head and body position on dizziness (physical
vestibular dysfunction (7,16,19).’’ manifestations), the Emotional Impact of dizziness
Most recently, a factor analytic investigation of the on everyday life and the effect of dizziness on the
DHI was reported by Ardic et al. (23). The investigators Catastrophic impact of dizziness and vertigo.
reported their findings drawn from a sample of 2,111 The dominance of the general factor means that
patients. Two subscales were proposed. The first factor clinicians and researchers need only compute and
was similar in content to the original functional and interpret one dizziness score. The three subscales
physical subscales. The second correlated with the emo- (i.e., the physical, emotional, and functional sub-
tional subscale. The analysis also confirmed the strong scales) developed empirically by Jacobson and New-
internal consistency reliability of the total score (i.e., the man (1) were not supported in our analyses. Instead,
general factor). These findings, and the recommendation we found that 11 items measured three group factors
to abandon any subscale structure, have been supported as well as a general factor, and 14 items measured
by several authors (14,16,19). only the general factor. Our results do not indicate that
In the present investigation we have demonstrated the subscales must be reported to reliably assess
validity for the creation of subscales that are designed dizziness. However, our results on the value added
to predict the presence of benign paroxysmal positional by the group factor scores (i.e., PRMSE) do support
vertigo (BPPV). This contention was first reported by reporting separate scores. That is, should clinicians
Whitney et al. (33) who extracted from the original 25- and researchers prefer a more detailed analysis of the
item DHI both five-item (P1, F5, P11, P13, and P25) and data then calculating responses for the Physical man-
two-item (F5 and P13) subscales. The investigators ifestations of dizziness and vertigo, Catastrophic
examined the predictive accuracy of responses of the impact of dizziness and vertigo, and the Emotional
subscales to the presence of BPPV. Using the hypothe- impact of dizziness and vertigo is recommended.
sized subscales, a diagnosis of BPPV for these patients
could be predicted successfully. That is, their results CONCLUSION
showed the probability of having BPPV increased sig-
nificantly as the patients endorsed a greater number of The results of the current investigation support two
items for both subscales. methods of analyzing data from the DHI: 1) continue
Similar to the a priori grouping of the items in the with the current practice of calculating a total score as the
original DHI, the items chosen by Whitney et al. (33) simple sum of the values for the 25 items (i.e., the total
were selected empirically and without statistical confir- DHI score that ranges from 0 to 100%), or, 2) calculate a
mation. Interestingly, analyses of several translated ver- total score and, then calculate subscale scores to assess
sions of the DHI were later revealed to contain subscales the Physical manifestations (i.e., sum of the items scores
that are identical to the proposed BPPV subscales. Both for 1, 5,11, 13, and 25), Catastrophic impact (i.e., sum of
our confirmatory analysis in the present investigation and the item scores for 9, 16, and 20), and Emotional Impact
the PCA of the Dutch (7) and German (20) versions (i.e., sum of the items scores for 2, 22, and 23) of
contain subscales with items identical to those in our dizziness on the dizzy patient.
five-item Physical Manifestations subscale. Likewise,
the analysis of the Japanese (13) version of the DHI REFERENCES
revealed five factors, one of which contained the same
items found in the 2-item subscale proposed by Whitney 1. Jacobson GP, Newman CW. The development of the Dizziness
et al. (33). Handicap Inventory. Arch Otolaryngol Head Neck Surg
Chen et al. (34) conducted an investigation to explore 1990;116:424–7.
2. Alsanosi AA. Adaptation of the Dizziness Handicap Inventory for
the validity and screening potential of the proposed use in the Arab population [in Arabic]. Neurosciences (Rihadh)
BPPV subscales in a Chinese population. Results of their 2012;17:139–44.
analyses revealed acceptable internal consistency reli- 3. Caldara B, Asenzo AI, Paglia GB, et al. Cross-cultural adaptation
ability with good predictive accuracy. Both the subscale and validation of the Dizziness Handicap Inventory: Argentine
version. Acta Otorrinolaringol Esp 2011;63:106–14.
scores as well as the total score significantly increased in 4. Castro AS, Gazzola JM, Natour J, Ganança FF. Brazilian version of
individuals diagnosed with BPPV. Previous research (35) the Dizziness Handicap Inventory. Pro Fono 2007;19:97–104.
has suggested the total DHI score is useful in predicting 5. Georgieva-Zhostova S, Kolev OI, Stambolieva K. Translation, cross-
BPP, a finding that is at odds with the data reported by cultural adaptation and validation of the Bulgarian version of the
Whitney et al. (33). The work conducted by Whitney Dizziness Handicap Inventory. Qual Life Res 2014;23:2103–7.
6. Poon DM, Chow LC, Hui Y, Au DK, Hui Y, Leung MC. Translation
et al. (33) and Chen et al. (34) both support the contention of the Dizziness Handicap Inventory into Chinese, validation of it,
that the ‘‘new’’ DHI subscales offered an accurate and evaluation of the quality of life of patients with chronic
screening tool to predict the presence of BPPV. dizziness. Ann Oto Rhinol Laryngol 2004;113:1006–11.
Copyright © 2019 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.
THE COMPONENT STRUCTURE OF THE DIZZINESS HANDICAP INVENTORY 1223
7. Vereeck L, Truijen S, Wuyts FL, Van De Heyning PH. Internal 21. Takano NA, Cavalli SS, Ganança MM, et al. Quality of life in
consistency and factor analysis of the Dutch version of the Dizzi- elderly with dizziness. Braz J Otorhinolaryngol 2010;76:769–75.
ness Handicap Inventory. Acta Otolaryngol 2007;127:788–95. 22. Zhang Y, Liu B, Wang Y, et al. Analysis of reliability and validity of
8. Kaplan DM, Friger M, Racover NK, Peleg A, Kraus M, Puterman the Chinese Version of Dizziness Handicap Inventory (DHI) [in
M. The Hebrew Dizziness Handicap Inventory [in Hebrew]. Hare- Chinese]. Zhonghua Er Bi Yan Hou Tou Jing Wai Ke Za Zhi
fuah 2010;149:697–700. 2015;50:738–43.
9. Nyabenda A, Briart C, Deggouj N, Gersdorff M. Normative study 23. Ardic FN, Tumkaya F, Akdag B, Senol H. The subscales and short
and reliability of French version of the Dizziness Handicap Inven- forms of the Dizziness Handicap Inventory: are they useful for
tory. Ann Readapt Med Phys 2004;47:105–13. comparison of the patient groups? Disabil Rehabil 2017;39:2119–
10. Kurre A, van Gool CJ, Bastiaenen CH, Gloor-Juzi T, Straumann D, 22.
de Bruin ED. Translation, cross-cultural adaptation and reliability 24. Chen FF, West SG, Sousa KH. A comparison of bifactor and
of the German version of the Dizziness Handicap Inventory. Otol second-order models of quality of life. Multivariate Behav Res
Neurotol 2009;30:359–67. 2006;41:189–225.
11. Nikitas C, Kikidis D, Katsinis S, Kyrodimos E, Bibas A. Translation 25. Reise SP, Moore TM, Haviland MG. Bifactor models and rotations:
and validation of the Dizziness Handicap Inventory in Greek exploring the extent to which multidimensional data yield univocal
language. Int J Audiol 2017;56:936–41. scale scores. J Pers Assess 2010;92:544–59.
12. Nola G, Mostardini C, Salvi C, Ercolani AP, Ralli G. Validity of 26. Jennrich RI, Bentler PM. Exploratory bi-factor analysis. Psycho-
Italian adaptation of the Dizziness Handicap Inventory (DHI) and metrika 2011;76:537–49.
evaluation of the quality of life in patients with acute dizziness. Acta 27. Yost KJ, Waller NG, Lee MK, Vincent A. The PROMIS fatigue
Otorhinolaryngol Ital 2010;30:190. item bank has good measurement properties in patients with
13. Goto F, Tsutsumi T, Ogawa K. The Japanese version of the fibromyalgia and severe fatigue. Qual Life Res 2017;26:1417–26.
Dizziness Handicap Inventory as an index of treatment success: 28. Reise SP. Invited paper: the rediscovery of bifactor measurement
exploratory factor analysis. Acta Olaryngol 2011;131:817–25. models. Multivariate Behav Res 2012;47:667–96.
14. Tamber AL, Wilhelmsen KT, Strand LI. Measurement properties of 29. Revelle W, Zinbarg RE. Coefficients alpha, beta, omega and the
the Dizziness Handicap Inventory by cross-sectional and longitu- glb: comments on Sijtsma. Psychometrika 2009;74:145–54.
dinal designs. Health Qual Life Outcomes 2009;7:101. 30. Reise SP, Scheines R, Widaman KF, Haviland MG. Multidimen-
15. Jafarzadeh S, Bahrami E, Pourbakht A, Jalaie S, Daneshi A. sionality and structural coefficient bias in structural equation
Validity and reliability of the Persian version of the Dizziness modeling: a bifactor perspective. Educ Psychol Meas 2013;73:
Handicap Inventory. J Res Med Sci 2014;19:769–75. 5–26.
16. Perez N, Garmendia I, Garcia-Granero M, Martin E, Garcia-Tapia 31. Lai JS, Butt Z, Wagner L, et al. Evaluating the dimensionality of
R. Factor analysis and correlation between Dizziness Handicap perceived cognitive function. J Pain Symptom Manage 2009;37:
Inventory and dizziness characteristics and impact on quality of life 982–95.
scales. Acta Otolaryngol Suppl 2001;545:145–54. 32. Lai JS, Crane PK, Cella D. Factor analysis techniques for assessing
17. Jarlsäter S, Mattsson E. Test of reliability of the Dizziness Handicap sufficient unidimensionality of cancer related fatigue. Qual Life Res
Inventory and the Activities-specific Balance Confidence Scale for 2006;15:1179–90.
use in Sweden. Adv Physiother 2003;5:137–44. 33. Whitney SL, Marchetti GF, Morris LO. Usefulness of the Dizziness
18. Karapolat H, Eyigor S, Kirazli Y, Celebisoy N, Bilgen C, Kirazli T. Handicap Inventory in the screening of benign paroxysmal posi-
Reliability validity and sensitivity to change of Turkish Dizziness tional vertigo. Otol Neurotol 2005;26:1027–33.
Handicap Inventory (DHI) in patients with unilateral peripheral 34. Chen W, Shu L, Wang Q, et al. Validation of 5-item and 2-item
vestibular disease. Int Adv Otol 2009;5:237–45. questionnaires in Chinese version of Dizziness Handicap Inventory
19. Asmundson GJG, Stein MB, Ireland D. A factor analytic study of for screening objective benign paroxysmal positional vertigo. Neu-
the Dizziness Handicap Inventory: does it assess phobic avoidance rol Sci 2016;37:1241–6.
in vestibular referrals. J Vestib Res 1999;9:63–8. 35. Saxena A, Prabhakar MC. Performance of DHI score as a predictor
20. Kurre A, Bastiaenen CH, van Gool CJ, Gloor-Juzi T, de Bruin ED. of benign paroxysmal positional vertigo in geriatric patients with
Exploratory factor analysis of the Dizziness Handicap Inventory dizziness/vertigo: a cross-sectional study. PLoS One 2013;
(German version). BMC Ear Nose Throat Disord 2010;10:3. 8:e58106.
Copyright © 2019 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.