Department of Rheumatology and Institute of Physical Medicine, University Hospital Zurich, Switzerland

Maastricht University, school CAPHRI, Department of Epidemiology and Faculty of Health Medicine and Life Sciences, the Netherlands

Institute of Human Movement Sciences and Sport, ETH, Zurich, Switzerland

Interdisciplinary Center for Vertigo & Balance Disorders, Departments of ENT, Neurology & Psychiatry, University Hospital Zurich, Switzerland

Abstract

Background

The Dizziness Handicap Inventory (DHI) is a validated, self-report questionnaire which is widely used as an outcome measure. Previous studies supported the multidimensionality of the DHI, but not the original subscale structure. The objectives of this survey were to explore the dimensions of the Dizziness Handicap Inventory - German version, and to investigate the associations of the retained factors with items assessing functional disability and the Hospital Anxiety and Depression Scale (HADS). Secondly we aimed to explore the retained factors according to the International Classification of Functioning, Disability and Health (ICF).

Methods

Patients were recruited from a tertiary centre for vertigo, dizziness or balance disorders. They filled in two questionnaires: (1) The DHI assesses precipitating physical factors associated with dizziness/unsteadiness and functional/emotional consequences of symptoms. (2) The HADS assesses non-somatic symptoms of anxiety and depression. In addition, patients answered the third question of the University of California Los Angeles-Dizziness Questionnaire which covers the impact of dizziness and unsteadiness on everyday activities. Principal component analysis (PCA) was performed to explore the dimensions of the DHI. Associations were estimated by Spearman correlation coefficients.

Results

One hundred ninety-four patients with dizziness or unsteadiness associated with a vestibular disorder, mean age (standard deviation) of 50.6 (13.6) years, participated. Based on eigenvalues greater one respectively the scree plot we analysed diverse factor solutions. The 3-factor solution seems to be reliable, clinically relevant and can partly be explained with the ICF. It explains 49.2% of the variance. Factor 1 comprises the effect of dizziness and unsteadiness on emotion and participation, factor 2 informs about specific activities or effort provoking dizziness and unsteadiness, and factor 3 focuses on self-perceived walking ability in relation to contextual factors. The first factor correlates moderately with disability and the HADS (values ≥0.6). The second factor is comparable with the original physical subscale of the DHI and factors retained in previous studies.

Conclusions

The results of the present survey can not support the original subscale structure of the DHI. Therefore only the total scale should be used. We discuss a possible restructuring of the DHI.

Background

The Dizziness Handicap Inventory (DHI) is a validated, self-report questionnaire designed to evaluate the precipitating physical factors associated with dizziness and unsteadiness as well as the functional and emotional consequences of vestibular disease

Although Jacobson & Newman

The internal validity of the content domains suggested by Jacobson and Newman was investigated for the original English

The primary objective of this study was to investigate the internal validity of the subscale structure and to explore the dimensions of the German version of the Dizziness Handicap Inventory (DHI-G). Since the DHI assesses disability and feelings of anxiety and depression, the associations of (i) the DHI, (ii) the original subscales and (iii) the factors retained by the exploratory factor analysis with single items assessing functional disability and the Hospital Anxiety and Depression Scale (HADS)

Methods

Participants

Patients who had suffered for at least one month from vertigo, dizziness or unsteadiness were included in the study. Problems had to be associated with a vestibular disorder. Further inclusion criteria were the ability to walk, to independently manage about 50% of the daily tasks, and to understand and speak German. Exclusion criteria were dizziness or unsteadiness exclusively due to cardiopulmonary, musculoskeletal, neurological or psychic disorders

Procedure

In the period between July 2007 and May 2009, participants were recruited from the Interdisciplinary Center for Vertigo & Balance Disorders, Departments of ENT, Neurology & Psychiatry at the University Hospital Zurich. Patients were referred to the center primarily for diagnostic reasons. The diagnostic procedure consisted of a detailed clinical history, a complete neuro-otological bedside examination, laboratory tests, and MR imaging of the brain with special emphasis on brainstem, cerebellum and vestibulo-cochlear nerves. All patients who fulfilled the inclusion criteria and not the exclusion criteria were included in the study, if they gave written consent. The ethics committee of the Canton of Zurich approved the study, which was the continuation of a previous survey on the reliability and validity of the DHI-G

Measures

The DHI is a 25-item questionnaire that was designed to help patients rate their self-perception of disability from dizziness

The 14-item Hospital Anxiety and Depression Scale (HADS) assesses independently non-somatic symptoms of anxiety (HADS-A) and depression (HADS-D). Each item is rated with 0 - 3 points. Scores on the two subscales range from zero (no sign of anxiety or depression) to 21 (maximum level of anxiety or depression). 8 - 10 points indicate possible and > 10 points probable anxiety or depression

In addition, patients rated their functional disability in one global question as mild, moderate or severe. They also answered question no. 3 of the 5-item University of California Los Angeles Dizziness Questionnaire (UCLA-DQ)

Data Analysis

Baseline characteristics of the study population, such as mean age, gender distribution, the degree of disability, and neuro-otologic diagnoses were assessed. The distributions of DHI-G and HADS scores were statistically investigated.

To evaluate the different dimensions of the DHI a principal component analysis (PCA) was conducted. Before PCA was done, various assumptions on inter-correlations of the 25 DHI items were tested

We opted to extract factors with eigenvalues greater than 1 [Kaiser's criterion (K1)] and to repeat the PCA after inspection of the scree plot, a graph plotting each eigenvalue against the factor

We conducted the PCA with oblique rotation and interpreted primarily the pattern matrix

We estimated the associations of the factors and the original subscales with 1) the items assessing functional disability and 2) the HADS by calculating Spearman's correlation coefficients. Coefficients < 0.25 were considered to indicate weak associations, 0.26 - 0.50 fair, 0.51 - 0.75 moderate and ≥0.76 strong associations

The internal consistency of the retained factors was investigated by estimating Cronbach's alpha coefficients and corrected item-total correlations (CI-TCs).

The analyses were computed using the SPSS version 16.0 computer software.

Results

Patient characteristics

One hundred ninety-four patients with a mean (standard deviation) age of 50.6 (13.6) years were included. Characteristics of the study population are summarized in Table

Baseline characteristics of the study population

**Characteristics of the patients**

Age (mean [SD], range) (yr)

50.6 (13.6)

21 - 77

Sex (n, [%])

Male

74 (38.1)

Female

120 (61.9)

Groups of diagnosis (n [%])

UPVD

72 (37.1)

BPVD

16 (8.2)

CVD

71 (36.6)

Multisensory/multifactorial

35 (18.0)

Duration of dizziness or unsteadiness (n [%])

> 1 mo and maximum 6 mo

55 (28.4)

> 6 mo and maximum 12 mo

25 (12.9)

> 12 mo

114 (58.8)

Level of disability (n [%])

Little

52 (26.8)

Moderate

98 (50.5)

Severe

44 (22.7)

Limitation in activity respectively participation (UCLA; Question 3) (n [%])

No effect at all

14 (7.2)

Continuing out all activities but with allowance for the dizziness

36 (18.6)

Continuing most of the activities

78 (40.2)

Continuing some of the activities

49 (25.3)

Unable to continue any of the activities

17 (8.8)

DHI-G total scale^{a}

Mean (SD)

44.8 (22.2)

Median (range)

44 (0 - 93)

Functional subscale^{b}

Mean (SD)

16.7 (9.7)

Median (range)

18 (0 - 36)

Physical subscale^{c}

Mean (SD)

13.9 (7.0)

Median (range)

14 (0 - 28)

Emotional subscale^{b}

Mean (SD)

14.1 (8.8)

Median (range)

14 (0 - 36)

HADS^{d}

Mean (SD)

11.5 (7.7)

Median (range)

10 (0 - 33)

Anxiety subscale^{e}

Mean (SD)

6.3 (4.3)

Median (range)

6 (0 - 17)

Depression subscale^{e}

Mean (SD)

5.1 (4.2)

Median (range)

4 (0 - 18)

BPVD indicates bilateral peripheral vestibular dysfunction; CVD, central vestibular dysfunction; DHI-G, Dizziness Handicap Inventory - German version; HADS, Hospital Anxiety Depression Scale; multisensory/multifactorial causes of dizziness; SD, standard deviation; UCLA-DQ, University of California Los Angeles - Dizziness Questionnaire; UPVD, unilateral peripheral vestibular dysfunction;

^{a }Maximum score of the DHI-G: 100 points; higher scores mean more disability

^{b }Maximum scores of the functional and emotional subscale: 36 points

^{c }Maximum score of the physical subscale: 28 points

^{d }Maximum score of the HADS: 42 points; higher scores mean more anxiety or depression

^{e }Maximum scores of the anxiety and depression subscale: 21 points

Factor analysis

Exploring the correlation matrix proved the variables to be suited for a factor analysis. The Determinant resulted in a value of 7.62E-006 which is slightly under the recommended value. Bartlett's test was highly significant (

The K1-criterion resulted in a 7-factor solution explaining 66.8% of the variance. Because 4 of the 7 factors consisted of less than 4 variables this solution was not further investigated. The inspection of the scree plot indicated 4- and 3-factor solutions (Figure

The Screeplot is a graph plotting each eigenvalue against the factor

**The Screeplot is a graph plotting each eigenvalue against the factor**. The initial eigenvalues of factor 1, 2, 3 and 4 are 8.381, 2.447, 1.469 and 1.321. After oblique rotation the values are in the 4- factor solution 6.202, 4.040, 5.350 and 3.044 and in the 3- factor solution 6.035, 4.252 and 6.080.

The Four-factor solution

The 4-factor solution explained 54.5% of the variance. The investigation of the fit of the model resulted in 129 (43%) non-redundant residuals. Communalities after factor extraction showed values in the range of 0.5, with F7 (difficulties in reading) having the lowest value (0.247).

The first factor consists of 9 items, 5 of these with factor loadings greater than 0.6 (Table

**Linking the items of the DHI to ICF-categories**. This file represents the linking of each DHI item to ICF-labels and ICF-categories.

Click here for file

The four factor solution of the principal component analysis

**Factor 4.1**

**Factor 4.2**

**Factor 4.3**

**Factor 4.4**

**Item**

**Abbreviated item description**

**Effect of dizziness and unsteadiness on emotion and participation**

**Specific activities/movements provoking dizziness or unsteadiness**

**Contextual factors or effort provoking dizziness and unsteadiness**

**Dependence of others/fear**

**load**

**CI-TC**

**Load**

**CI-TC**

**load**

**CI-TC**

**load**

**CI-TC**

E 23

feeling depressed

**0.788**

0.68

E 2

feeling frustrated

**0.765**

0.60

F 24

job/house responsibilities

**0.685**

0.71

E 21

feeling handicapped

**0.663**

0.76

E 10

embarrassed in front of others

**0.623**

0.38

F 6

restriction of social activities

0.578

0.74

E 22

stressed relationships

0.559

0.65

F 3

restriction of travel

0.477

0.65

E 18

difficulties in concentrating

0.452

0.49

Cronbachs alpha

0.88

P 13

turning over in bed

**0.824**

0.56

F 5

getting into or out of bed

**0.751**

0.56

P 1

looking up

**0.672**

0.49

P 11

quick head movements

**0.656**

0.52

P 25

bending over

0.589

0.49

F 7

difficulties in reading

0.193

0.265

0.27

0.124

0.169

Cronbachs alpha

0.74

P 17

walking down a sidewalk

**0.738**

0.59

F 19

walking around in dark

**0.715**

0.44

E 15

afraid of appearing intoxicated

**0.603**

0.51

P 8

ambitious activities like sports

0.539

0.63

P 4

walking down a supermarket aisle

0.502

0.48

F 12

avoid heights

0.482

0.48

F 14

strenuous housework

0.412

0.64

Cronbachs alpha

0.80

E 20

afraid to stay home alone

**0.759**

0.52

E 9

afraid of leaving home alone

**0.707**

0.70

F 16

walking by yourself

**0.512**

0.62

Cronbachs alpha

0.77

The table indicates factor loadings, corrected item-total correlations (CI-TC) and Cronbachs alpha correlation coefficients estimated in n = 194. Values of factor loadings are results of the pattern matrix. Bold face indicates loadings with absolute values of 0.6 or more. Loadings < 0.4 are not reported with exception of the values of F7, which has only low loadings. The 4-factor solution explained 54.5% of the variance.

Abbreviations: E indicates emotional subscale; F, functional subscale; P, physical subscale of the Dizziness Handicap Inventory - German version; load indicates factor loading.

Associations between the DHI, the original subscales, the retained factors, disability items and the HADS

**Level of disability**

**Limitation of daily activity/participation**

**(UCLA-DQ3)**

**HADS**

**HADS-A**

**HADS-D**

DHI-G

**0.71****

**0.55****

**0.59****

0.43**

**0.66****

E

**0.63****

**0.57****

**0.62****

0.46**

**0.67****

F

**0.71****

**0.51****

**0.54****

0.39**

**0.62****

P

0.46**

0.34**

0.36**

0.26**

0.40**

Factor

3.1

**0.70****

**0.60****

**0.60****

0.43**

**0.68****

Factor

3.2

**0.53****

0.40**

0.35**

0.26**

0.39**

Factor

3.3

**0.53****

0.34**

**0.52****

0.39**

**0.56****

Factor

4.1

**0.70****

**0.60****

**0.60****

0.43**

**0.68****

Factor

4.2

0.44**

0.35**

0.27**

0.22**

0.28**

Factor

4.3

**0.53****

0.33**

0.47**

0.33**

**0.55****

Factor

4.4

**0.52****

0.42**

0.47**

0.37**

0.48**

Values are Spearman correlation coefficients: ** correlation is significant at 0.01 level (1-tailed); * correlation is significant at the 0.05 level (1-tailed). Bold face indicates moderate associations.

Abbreviations: DHI-G indicates Dizziness Handicap Inventory-German version; E, F and P, emotional, functional, and physical subscales of the DHI-G; Factor 3.1 - 3.3 indicate the 3 components of the 3-factor solution; Factor 4.1 - 4.4 the 4 components of the 4-factor solution; HADS, Hospital Anxiety and Depression Subscale; HADS-A, anxiety subscale of the HADS; HADS-D, depression subscale of the HADS; UCLA-DQ3, item 3 of the University of California Los Angeles -Dizziness Questionnaire

The second factor is composed of 6 variables, 4 of these with factor loadings > 0.6 (Table

The third factor comprises of 7 items, 3 of these with factor loadings > 0.6. This factor assesses how contextual factors or effort relate to dizziness, unsteadiness and self-perceived walking ability. Depending on the ability to process and differentially use afferent input, individuals may feel dizzy or unsteady when walking down a sidewalk (P17), walking around in the dark (F19), walking down a supermarket aisle (P4), or being in height (F12). In our opinion item E15, 'afraid of appearing intoxicated', also targets the self-perceived walking ability. Many patients who can not walk straight ahead affirm this question. According to the linking rules as described by Cieza et al. (2005) item P8 (ambitious activities like sport) and F14 (strenuous housework) target participation (sport: ICF-category d0201; doing housework: ICF-category d640) as well as body function (muscle power function: ICF category b730

The fourth factor consists of 3 items, 2 with factor loadings > 0.6 (Table

The correlation coefficients show marginally moderate associations between the third and fourth factor and the self estimated level of disability (r = 0.53 and 0.52) and between the third factor and HADS-D (0.55) (Table

The three factor solution

The 3-factor solution explained 49.2% of the variance. The investigation of the fit of the model resulted in 137 (45%) non-redundant residuals. The inspection of the communalities showed values in the range of 0.5, with F7 (difficulties in reading) and E20 (afraid to stay home alone) having the lowest values (0.24, 0.26). The first factor is composed of the same items as factor 1 of the 4-factor solution (Table

The three factor solution of the principal component analysis

**Factor 3.1:**

**Factor 3.2:**

**Factor 3.3:**

**Item**

**Abbreviated item description**

**Item mean (SD)**

**Effect of dizziness and unsteadiness on emotion and participation**

**Specific activities/movements or effort provoking dizziness or unsteadiness**

**self-perceived**

**walking ability and feeling of postural stability in relation to contextual factors**

**load**

**CI-TC**

**load**

**CI-TC**

**load**

**CI-TC**

E 23

feeling depressed

1.8 (1.5)

**0.792**

0.68

E 2

feeling frustrated

2.4 (1.4)

**0.757**

0.59

E 21

feeling handicapped

2.3 (1.5)

**0.653**

0.76

F 24

job/house responsibilities

2.0 (1.6)

**0.652**

0.71

E 10

embarrassed in front of others

1.6 (1.9)

**0.622**

0.38

F 6

restriction of social activities

2.1 (1.7)

0.567

0.74

E 22

stressed relationships

1.3 (1.6)

0.563

0.65

F 3

restriction of travel

2.2 (1.7)

0.463

0.65

E 18

difficulties in concentrating

1.9 (1.5)

0.397

0.49

Cronbachs alpha

0.88

P 13

turning over in bed

1.7 (1.6)

**0.747**

0.52

P 11

quick head movements

2.9 (1.5)

**0.726**

0.55

P 1

looking up

2.3 (1.6)

**0.707**

0.48

F 5

getting into or out of bed

1.6 (1.6)

**0.683**

0.56

P 25

bending over

2.0 (1.7)

**0.618**

0.48

P 8

ambitious activities like sports

2.3 (1.7)

0.439

0.56

F 14

strenuous housework

2.1 (1.6)

0.410

0.60

F 7

difficulties in reading

1.5 (1.6)

0.200

0.253

0.35

0.209

Cronbachs alpha

0.80

P 17

walking down a sidewalk

1.1 (1.5)

**0.713**

0.58

F 19

walking around in dark

1.7 (1.7)

**0.701**

0.42

F 16

walking by yourself

1.2 (1.5)

**0.623**

0.61

E 15

afraid of appearing intoxicated

1.4 (1.7)

0.580

0.51

E 9

afraid of leaving home alone

1.0 (1.3)

0.572

0.58

P 4

walking down a supermarket aisle

1.8 (1.7)

0.553

0.51

F 12

avoid heights

2.3 (1.8)

0.543

0.43

E 20

afraid to stay home alone

0.5 (1.1)

0.416

0.44

Cronbachs alpha

0.79

The table indicates item statistics, factor loadings, corrected item-total correlations (CI-TC) and Cronbachs alpha correlation coefficients estimated in n = 194. Values of factor loadings are results of the pattern matrix. Bold face indicates loadings with absolute values of 0.6 or more. Loadings < 0.4 are not reported with exception of the values of F7, which has only low loadings. The 3-factor solution explained 49.2% of the variance.

Abbreviations: E indicates emotional subscale; F, functional subscale; P, physical subscale of the Dizziness Handicap Inventory - German version; load indicates factor loading.

Like in the 4-factor solution, the first factor shows moderate associations with the items assessing functional disability (

The 3 - factor solution seems to be the most reliable solution and holds clinical relevant dimensions. Cronbach alpha coefficients of the retained factors and corrected item-total correlations (CI-TCs) within each factor fulfil the commonly accepted minimal standards of 0.7 for Cronbachs alpha and 0.2 for CI-TCs

Discussion

The exploratory factor analysis of the German version of the Dizziness Handicap Inventory (DHI-G) led to a clinically interesting 3-factor solution which seems to be reliable with at least 8 variables per factor and twice 5- and once 3-factor loadings greater than 0.6. Although the three factors differ from the original 3 subscales, the objectives of the DHI to quantify the functional and emotional consequences of dizziness or imbalance as well as to assess symptom provoking activities are supported. While factor 1 assesses the effect of dizziness and unsteadiness on emotion and participation, factor 2 informs about specific activities/movements or effort provoking dizziness or unsteadiness, and factor 3 about self-perceived walking ability and the feeling of postural stability in relation to contextual factors. The dimension as targeted by factor 1 is supported by moderate associations with the Hospital Anxiety and Depression Scale and items assessing functional disability. The found 3-factor solution is clinically interesting. While the first factor might indicate whether health care professionals, such as social workers, occupational therapists or psychologists, should become involved in the assessment and treatment of the individual, the scores of the second and third factors might indicate whether a patient will benefit from therapy, primarily emphasizing physical or behavioural training.

The factor analyses of the English

**Comparison of the 3-factor solution among different factor analysis studies of the Dizziness Handicap Inventory**. This file represents the results of the 3-factor solutions of the English, Spanish, Dutch, and German version of the DHI.

Click here for file

Factor 2 is the most similar among all language versions. The number of enclosed items ranges between 2 and 8. P13 (turning over in bed) and F5 (getting into or out of bed) are included in all second factors. P11 (quick head movements), P25 (bending over) and P8 (ambitious activities like sports) are part of the second factor in three; P1 (looking up) and F14 (strenuous housework) are included in two language versions. Authors describe the dimension assessed by this factor primarily as limitations in specific activities in relation to motion sensitivity.

The common objective targeted by the third factor is the aspect of visuo-vestibular dysfunction and context dependent behaviour. P4 (walking down the supermarket aisle) is the item included in the third factor of all language versions. P17 (walking down a sidewalk), F19 (walking around in the dark), E15 (afraid of appearing intoxicated) and F12 (avoid heights) belong to this factor in three of the language versions. F16 (walking by yourself), E9 (afraid of leaving home alone) and E20 (afraid to stay home alone), all assessing the dependence of personal assistance, belong to factor 3 of the English and German version, and to factor 1 of the Dutch and Spanish version.

The differences in the results of the various factor analysis studies may be attributed to several factors. One important aspect is the study population. The sample sizes vary from 95

The most critical point and therefore a limitation of our study can be attributed to the nature of PCA. For PCA, one assumes that variables are numeric and normally distributed. Items of the DHI, however, are ordinal. We therefore repeated the factor analysis with a Categorical Principal Component Analysis (CATPCA) restricted to 4 and 3 factors. In both cases, the analysis resulted in a quasi 2-factor solution, with nearly all items - except P13 (turning over in bed), P11 (quick head movements), P1 (looking up), F5 (getting into or out of bed) and P25 (bending over) - in factor 1. This result supported the stability of a dimension assessing "motion sensitivity" represented by these 5 items. Disregarding F7 (difficulties in reading), these items encompass our second factor of the 4-factor solution (Table

The interpretation of the dimensions of the DHI was mainly done by identifying the keywords of the questions and linking them with the ICF components respectively categories

Jacobson & Newman

Future research should further investigate the construct validity of the newly defined dimensions of the DHI. It could be hypothesized that factor 3 moderately correlates with tests assessing walking ability e.g. the Dynamic Gait Index

Conclusions

The Dizziness Handicap Questionnaire is a disease-specific health-related quality of life questionnaire. Like in previous studies the original subscale structure could not be supported, but the multidimensionality was obvious. The found 3-factor-solution showed comparable aspects with the results of previous factor analysis studies of the DHI. The retained factors could partly be interpreted with the ICF. The construct of the first factor could be supported by moderate associations with functional disability and non-somatic symptoms of anxiety and depression. In our opinion the 3 retained factors seem to be helpful for diagnostic or interventional decisions. Therefore a restructuring of the DHI might be discussed.

Competing interests

The authors declare that they have no competing interests.

Authors' contributions

AK contributed to the design of the survey. She conducted the statistical analysis and wrote the manuscript. CHGB contributed to the analysis of data and revised the article critically for its content. CJAWvG attributed to the design of the study, contributed to the interpretation of data and revised the article critically for its content. TG-J contributed to the design of the study, the acquisition of data, the interpretation of data and revised the article critically for its content. EDdB contributed to the analysis of data and revised the article critically for its content. DS co-initiated the study, contributed to the interpretation of data and revised the article critically for its content. All authors read and approved the final manuscript.

Acknowledgements

The authors wish to thank the patients for their interest and collaboration while answering all the questions, and S. Hegemann, E. Buffone, L. Rasi, S. Wittwer for their help in recruiting patients.

Pre-publication history

The pre-publication history for this paper can be accessed here: