Exploratory Factor Analysis of the Vestibular Activities Avoidance Instrument

Key Points

Question

Can a shortened version of the Vestibular Activities Avoidance Instrument (VAAI) with acceptable psychometric properties identify fear avoidance beliefs among persons with dizziness?

Findings

In this cohort study of 404 adults with dizziness, the VAAI was reduced from 81 items to 9 items (VAAI-9) by using exploratory factor analysis. The VAAI-9 had excellent internal consistency and was associated with activity limitations and participation restrictions, quality of life, and anxiety and depression symptoms.

Meaning

This cohort study found that the VAAI-9 showed strong psychometric properties, suggesting that it could be used to identify fear avoidance beliefs, which are important to measure in clinical settings owing to their association with disability and quality of life in persons with dizziness.

Abstract

Importance

Fear avoidance is a behavioral response to dizziness that can lead to chronic symptoms and maladaptation of the vestibular system, but there is no valid and reliable clinical measure of fear avoidance for persons with dizziness. Although the Vestibular Activities Avoidance Instrument (VAAI) was developed to identify fear avoidance beliefs in persons with dizziness, it was considered too long for clinical use.

Objective

To continue development of the VAAI for clinical use by reducing its length and by assessing the internal consistency and construct validity through associations with measures of disability, quality of life, and psychological well-being.

Design, Setting, and Participants

This prospective cohort study conducted from February 2018 to December 2019 at a tertiary care balance disorders clinic and in outpatient physical therapy clinics in the United States included 404 adults with dizziness.

Main Outcomes and Measures

Participants completed the 81-item VAAI, the Vestibular Activities and Participation (VAP) measure, the 12-Item Short Form Health Survey (SF-12), and the Hospital Anxiety and Depression Scale (HADS) at baseline. Exploratory factor analysis of the VAAI was conducted to reduce the number of items. Internal consistency of the reduced VAAI was determined by calculating the Cronbach α. Convergent validity was assessed by examining the associations between the reduced VAAI and the VAP, the SF-12, and the HADS using Spearman correlation coefficients.

Results

Data from 404 adults (mean [SD] age, 54.0 [17.0] years; 64.6% women) were included in the analyses. The exploratory factor analysis indicated that 2 factors explained the underlying constructs of the 81-Item VAAI. The first factor was retained and measured the construct of fear avoidance. The VAAI was reduced to 9 items (VAAI-9). The VAAI-9 showed excellent internal consistency (Cronbach α = 0.92) and was positively associated with the VAP (ρ = 0.81) and the HADS anxiety (ρ = 0.47) and depression (ρ = 0.64) subscales, and negatively associated with physical (ρ = −0.76) and mental (ρ = −0.47) health–related quality of life.

Conclusions and Relevance

These findings indicate that the VAAI-9 is a short, internally consistent, valid measure of fear avoidance and is associated with quality of life, activity limitations and participation restrictions, and psychological well-being. The next steps in the development of the VAAI-9 will include validation in an external sample, assessment of test-retest validity, and prospective investigations of its association with future disability.

This cohort study applies exploratory factor analysis to the 81-item Vestibular Activities Avoidance Instrument to assess whether it can be shortened for practical clinical application to identify fear avoidance beliefs in individuals with dizziness.

Introduction

Individuals with vestibular disorders have increased prevalence of anxiety and depression when compared with the general population.^{1,2,3,4,5,6,7} The addition of comorbid psychiatric conditions is associated with increased symptom burden, poorer quality of life, and longer recovery time.^8,9,10,11,12 Persons experiencing dizziness often avoid activities and are unable to perform activities of daily living and required work duties.¹³ The fear and avoidance of movement and activities are associated with prolonged symptoms and may inhibit necessary vestibular compensation mechanisms.^12,14,15 However, fear avoidance and other behavioral and psychological factors are not often measured clinically because there are few tools available to measure these constructs in persons with vestibular disorders.

The Vestibular Activities Avoidance Instrument (VAAI) was developed with the goal to provide clinicians with a comprehensive patient-reported outcome measure that can identify fear avoidance and other psychological factors that may be associated with treatment response and outcomes exhibited by persons with vestibular disorders.¹⁶ The authors developed the VAAI by using reliable and valid outcome measures designed to measure fear avoidance, anxiety, somatization, depression, worry, and other psychological factors among persons with pain and adapted the items for use in persons with dizziness.^{17,18,19,20,21,22,23,24,25} Owing to the length of the original 81-Item VAAI, its clinical usefulness was considered minimal because of respondent burden.

Exploratory factor analysis (EFA) is commonly used in biomedical research to explain the associations among a set of items or indicators.²⁶ The assumption of EFA is that there are a small number of underlying factors that can explain the associations among the items.²⁷ The original VAAI included numerous psychological and behavioral constructs that may contribute to disability in persons with vestibular disorders. However, its factor structure has not been evaluated, and no efforts have yet been undertaken to reduce and optimize its content. Therefore, the aims of this study were to (1) conduct EFA to determine the internal constructs of the VAAI as a guide to reducing the number of items, (2) evaluate the internal consistency of a modified VAAI, and (3) evaluate for construct validity of the modified VAAI. We hypothesized that the VAAI would be factorable and that a modified version of the VAAI with fewer items would show adequate internal consistency and evidence of construct validity through significant associations with measures of disability, health-related quality of life, and psychological well-being.

Methods

Participants and Setting

This prospective cohort study included a baseline and 3-month follow-up assessment. For the purposes of this analysis, the data from the baseline assessment only were used. Participants were recruited from a tertiary care balance disorders clinic and outpatient physical therapy clinics from February 2018 to December 2019. Eligibility criteria included the following: currently experiencing dizziness, aged 18 to 100 years, and English speaking. This study was approved by the University of Pittsburgh institutional review board. All participants provided written informed consent to participate in the study and completed the in-person baseline assessment via computerized questionnaire. Individuals who agreed to participate received compensation for participating in this study.

Outcome Measures

Participants completed the 81-item VAAI, the Hospital Anxiety and Depression Scale (HADS), a modified version of the 12-Item Short Form Health Survey (SF-12), and the Vestibular Activities and Participation (VAP) measure at the baseline visit. The original version of the VAAI was 77 items but was modified by adding 4 additional items for the present study.¹⁶ The VAAI includes questions from other measures of psychological and behavioral factors, adapted for use in a population with unsteadiness, dizziness, or both.^{17,18,19,20,21,22,23,24,25} All items except 1 item are scored on a scale from 0 to 6, with 0 indicating “strongly disagree” and 6 indicating “strongly agree.” The last item differs from the other items and is scored on a scale from 0 to 4. Therefore, the total possible score ranges from 0 to 484, with higher scores indicating higher levels of fear avoidance beliefs and other psychological symptoms.

The HADS is commonly used to identify anxiety and depression in clinical settings.^28,29 Because the VAAI and HADS include theoretically related constructs (fear avoidance, anxiety, and depression), it was hypothesized that these 2 scales were significantly correlated. The SF-12 was designed to measure health-related quality of life and is positively associated with Dizziness Handicap Inventory (DHI) scores among women with peripheral vestibular disorders.^30,31 The VAP was developed specifically to measure activity limitations and participation restrictions that may be present in persons with vestibular disorders.³² Items were developed using the International Classification of Functioning, Disability, and Health, and the scale ranges from 0 to 4. It was hypothesized that greater activity and participation restrictions measured by the VAP and lower quality of life measured by the SF-12 would be associated with higher levels of fear avoidance beliefs measured by the VAAI.

Statistical Analysis

It is recommended that the minimum number of participants required for exploratory factor analysis is 5 participants per 1 item.³³ Therefore, a sample size of 405 was considered adequate to perform exploratory factor analysis on the 81-item VAAI.

Exploratory Factor Analysis

Descriptive statistics were calculated for each item on the VAAI and the total score. The Bartlett test of sphericity and the Kaiser-Meyer-Olkin test were conducted to determine if there were a sufficient number of significantly correlated items.^34,35 The individual measures of sampling adequacy for each item were evaluated, and items with values lower than 0.60 were removed because these items were not adequately correlated with other items in the scale.³⁵

After determining adequacy of sample size and relatedness of items using the tests above, the EFA was conducted using principal component analysis. Guidance regarding the number of factors to retain was derived by visualizing the scree plot and by identifying factors that explained greater than 5% of the variance.³⁵ The scree plot is a graph of the number of factors plotted against their eigenvalues. The inflection point of the slope marks the number of factors with meaningful contributions to variance. Factors that explain less than 5% of variance typically do not add useful information to a questionnaire. The number of factors suggested by each method was compared to determine the final number of factors to retain.^35,36 The factor structure was rotated obliquely using direct oblimin and promax procedures.^37,38 Oblique rotation was chosen because the factors were expected to be partially correlated.³⁵ All analyses were completed using IBM SPSS Statistics, version 25.

Reliability and Item Reduction

Internal consistency was calculated using Cronbach α. Item reduction was completed by including items with the highest factor loadings and then evaluating internal consistency with Cronbach α. This procedure was repeated multiple times by adding the next 2 highest loaded items at a time, and the Cronbach α was plotted against the number of items for each iteration.³⁵ The goal was to determine the point at which the addition of more items did not substantially improve internal consistency. Cronbach α values between 0.7 and 0.95 were considered evidence for internal consistency without redundancy among items for the final version of the VAAI.³⁹

Construct Validity

Construct validity was determined by taking numerous random samples of the original sample and examining the correlation between the shortened VAAI total score and the HADS, the SF-12, and the VAP using Spearman correlation coefficients and the bootstrapping resampling method.^40,41 The bootstrapping method took 1000 random samples of a specific size from the existing sample of data, where each data point was replaced after sampling. Although the bootstrapping method was used to provide a more accurate estimate of external validity for the shortened version of the VAAI, external validity should be assessed in a separate sample in the future.⁴² A correlation coefficient of greater than 0.3 was considered adequate to determine convergent validity.⁴³

Results

Demographic Characteristics

The sample included 404 participants reporting dizziness or imbalance symptoms. The mean (SD) age was 54.0 (17.0) years, and 64.6% of the sample were women (Table 1). The median (interquartile range) duration of dizziness was 8 (3-24) months and duration of imbalance was 9 (3-24) months.

Table 1. Demographic and Clinical Characteristics of 404 Individuals With Dizziness.

Variable	Mean (SD) Value
Age, y	54.0 (17.0)
Female, No. (%)	261 (64.6)
Duration, mo
Dizzinessa	42.4 (104.0)
Imbalanceb	38.1 (95.7)
Outcome measure (score range)
VAAI-9 (0-54)	25.3 (14.0)
VAP (0-4)	1.1 (0.8)
HADS-A (0-21)	7.2 (4.4)
HADS-D (0-21)	5.5 (4.1)
SF-12 PCS (0-100)	39.5 (10.7)
SF-12 MCS (0-100)	45.4 (11.1)

Abbreviations: HADS-A, Hospital Anxiety and Depression Scale–Anxiety Subscale; HADS-D, Hospital Anxiety and Depression Scale–Depression Subscale; MCS, Mental Component Summary; PCS, Physical Component Summary; SF-12, 12-Item Short Form Health Survey; VAAI-9, Vestibular Activities Avoidance Instrument 9-Item; VAP, Vestibular Activities and Participation.

^aMedian (interquartile range) duration of dizziness, 8 (3-24) months.

^bMedian (interquartile range) duration of imbalance, 9 (3-24) months.

Exploratory Factor Analysis

The result of the Bartlett test of sphericity was χ²₃₁₆₀ = 21099.24, and the Kaiser-Meyer-Olkin value was 0.95, indicating that there were a sufficient number of correlated items in the matrix to conduct the EFA. All individual measures of sampling adequacy values were greater than 0.60, except for 5 items. These were the 5 reverse-coded items. Because the measures of sampling adequacy values were less than 0.60, all 5 items were removed from the analysis. The Bartlett test of sphericity and the Kaiser-Meyer-Olkin test were repeated with similar results.

Two factors accounted for greater than 5% of the variance in the scale. The inflection in the scree plot occurred at 3 factors. Therefore, principal component analysis and factor rotation with direct oblimin and promax procedures were conducted for both 2- and 3-factor solutions. The 2-factor solution offered a cleaner result that was easier to interpret. In the 3-factor solution, only 6 of 81 items loaded on the third factor. Furthermore, the 3-factor solution produced greater diffusion of item loadings across factors than the 2-factor solution, which more clearly separated items. The direct oblimin and promax rotations produced similar solutions. Factor 1 included 40 items after direct oblimin rotation and appeared to measure the construct of activity and participation avoidance. Factor 2 included 23 items using the direct oblimin rotation and appeared to measure the construct of anxiety and worry.

Item Reduction and Reliability

Factor 1 contained items assessing the primary construct of interest, namely avoidance of activities and restriction of participation. Ten items from factor 1 with loadings less than 0.5 were eliminated.^44,45,46 These items tended to have modest loadings on factor 1, indicating a lack of specificity for the construct of interest. All items from factor 2 were eliminated because they nearly replicated a simple combination of questions from long-established, well-validated anxiety scales. For example, factor 2 included 5 of the 7 items from the Generalized Anxiety Disorder–7, and 9 of 14 items from the Short Health Anxiety Inventory. Given that the aim of this study was to construct a new activities avoidance instrument, there was little to be gained by retaining a large subset of items that nearly replicated existing measures. An EFA was conducted for the remaining 29 items in factor 1 and confirmed that these items loaded on 1 factor only. Two items were removed because they were strongly correlated (Pearson r > 0.7) with other items in the scale.

The remaining 27 items were then categorized into conceptual groups, including activities and participation, fear, work, and functional movement. On further review of the 5 items included in the “functional movement” category, the items were eliminated because they (1) had the lowest factor loadings, (2) did not capture the construct of activity avoidance, and (3) lowered internal consistency of the scale when they were included. After these 5 items were removed, the VAAI included 22 items measuring 3 concepts: activity and participation, fear, and work (Figure 1).

Figure 1. Conceptual Groupings in Factor 1 of the Vestibular Activities Avoidance Instrument.

Factor 1 contains 22 items relating to 3 concepts of activity avoidance.

The final form of the revised VAAI was constructed by choosing the item with the highest factor loading from each of the 3 remaining conceptual groups and calculating Cronbach α. The process was repeated until all 22 items were included. A plot of the α coefficients on the y-axis and number of items on the x-axis showed that a 9-item scale had excellent internal consistency (Cronbach α = 0.92) and did not improve to a large degree with the addition of more items (Figure 2). Therefore, a revised version of the VAAI was set at the 9 items (VAAI-9), which included 3 items from each of the conceptual groups (Table 2). The total score is an aggregate of the item scores and ranges from 0 to 54, with higher scores indicating more fear avoidance beliefs.

Figure 2. Cronbach α Values by Number of Items in Factor 1 of the Vestibular Activities Avoidance Instrument.

The point at which the addition of more items does not substantially increase Cronbach α is a 9-item scale.

Table 2. Vestibular Activities Avoidance Instrument 9-Item Factor Loadings and Internal Consistency.

Item No.	Item	Factor loading	Corrected item-total correlation	Cronbach α with item deleted
1	It is difficult for me to do strenuous homework or yard work because of my dizziness.	0.809	0.785	0.899
2	My participation in social activities, such as going out to dinner, going to the movies, dancing, or going to parties, is significantly restricted because of my dizziness.	0.795	0.772	0.900
3	My dizziness interferes with my job or household responsibilities.	0.786	0.779	0.900
4	I cannot do physical activities, which might make my dizziness worse.	0.771	0.713	0.904
5	I can’t do all the things normal people do because of my dizziness.	0.763	0.735	0.903
6	I am afraid that I might make myself dizzy or unsteady if I exercise.	0.719	0.654	0.908
7	I am afraid to leave my home without having someone go with me because of my dizziness.	0.685	0.610	0.911
8	I should not do my regular work with my present dizziness.	0.678	0.645	0.909
9	My work makes my dizziness worse.	0.670	0.618	0.910

Convergent Validity

The VAAI-9 and VAP scores at baseline were correlated (ρ = 0.81), indicating that greater fear avoidance was associated with greater activity limitations and participation restrictions (Table 3). The VAAI-9 showed a negative association with the SF-12 physical component summary score (PCS) (ρ = −0.76) and a negative association with the SF-12 mental component summary score (ρ = −0.47), indicating greater fear avoidance was associated with poorer physical and mental health-related quality of life. The VAAI-9 and the HADS anxiety (ρ = 0.47) and depression (ρ = 0.64) subscales were also correlated, indicating that activity avoidance measured by the VAAI-9 was associated with anxiety and depression symptoms. All these associations are consistent with the concept of fear avoidance, providing evidence of convergent validity for the VAAI-9.

Table 3. Correlations Between VAAI-9 and VAP, SF-12, and HADS Among 404 Adults With Dizziness.

Scale	VAAI-9 Spearman ρ	Bias-corrected (95% CI)
VAP	0.81	(0.77 to 0.84)
SF-12 PCS	−0.76	(−0.80 to −0.71)
SF-12 MCS	−0.47	(−0.54 to −0.38)
HADS-A	0.47	(0.38 to 0.54)
HADS-D	0.64	(0.58 to 0.69)

Abbreviations: HADS-A, Hospital Anxiety and Depression Scale–Anxiety Subscale; HADS-D, Hospital Anxiety and Depression Scale–Depression Subscale; MCS, Mental Component Summary score; PCS, Physical Component Summary; SF-12, 12-Item Short Form Health Survey; VAAI-9, Vestibular Activities Avoidance Instrument 9-Item; VAP, Vestibular Activities and Participation.

Discussion

The aim of this study was to develop a valid and internally consistent tool that is feasible to use in the clinic to measure the construct of fear avoidance in persons with dizziness. Through the process of EFA, item reduction, and internal consistency analysis, the 81-Item VAAI was organized into a short, internally consistent instrument including 9 items that can be used to measure fear avoidance in persons with dizziness in clinical settings. The VAAI-9 had a strong association with activity limitations and participation restrictions and moderate associations with anxiety and depression symptoms and quality of life among persons with vestibular disorders, demonstrating convergent validity.

There are few tools available that measure the construct of fear avoidance beliefs among persons with dizziness. The DHI has an emotional subscale component thought to identify psychological factors.¹⁷ The developers of the Vestibular Rehabilitation Benefits Questionnaire included 1 item to measure avoidance behavior with a total of 3 items to measure anxiety symptoms.⁴⁵ The Dizziness Catastrophizing Scale was developed by adapting the Pain Catastrophizing Scale for persons with dizziness and has shown internal consistency and validity in persons with vestibular disorders.⁴⁷ The Dizziness Catastrophizing Scale measures the cognitive construct of catastrophization rather than the behavioral concept of fear avoidance, which may be more germane to successful completion of rehabilitation and return to full participation in activities.

Through item reduction techniques using statistical analyses and expert opinion, the VAAI-9 was shortened from the original 81 items to the final 9 items, which showed excellent internal consistency without evidence of redundancy. The VAAI-9 contains items assessing individuals’ beliefs about their abilities to participate in job responsibilities, household chores, social events, physical activity, and exercise as well as their fears about making dizziness worse, engaging in exercise, and going outside the home.

The VAAI-9 showed a strong association with the VAP, indicating that activity limitations and participation restrictions are associated with fear and avoidance of activities, as hypothesized. This is not surprising given that some of the items contained within the VAAI-9 were adapted from the DHI, and the VAP has shown a strong association with the DHI in other studies.^32,48 The VAP measures limitations in activities and participation based on the International Classification of Functioning, Disability, and Health. These limitations may be due to physical symptoms of a vestibular disorder or from associated emotional symptoms. In contrast, the VAAI was developed to measure individuals’ convictions about illness and fears about participating in activities that may directly reduce their engagement in daily activities inside and outside the home, including necessary therapeutic interventions.

We hypothesized that the VAAI-9 would be significantly associated with quality of life and anxiety and depression symptoms. The results indicate that greater fear avoidance beliefs were strongly associated with poorer physical and moderately associated with poorer mental health-related quality of life. Previous studies have found significant negative associations between the DHI, VAP, and various measures of physical and mental health-related quality of life among persons with dizziness.^32,49,50 There is a well-studied association between psychiatric symptoms and vestibular disorders, indicating shared neural networks that link anxiety, fear, and dizziness.^51,52,53 There is also a higher prevalence of anxiety and depressive disorders among persons with dizziness when compared with the general population.^1,8,9 Previous research studies have found associations between dizziness handicap and psychiatric symptoms.^11,46,54 The results of the present study indicating a moderate association between the VAAI-9 and HADS scores align with the results of previous research and strengthens evidence for convergent validity.^55,56,57

Limitations

There are several limitations to the present study. First, the study used a convenience sample. The individuals who chose not to complete the study may have differed from those who did complete the study; thus, applicability to the investigators’ entire practices cannot be assured, and generalizability to other settings will have to be established in future investigations. Participants who were recruited from physical therapy clinics vs the tertiary care balance disorders clinic differed in some patient-reported outcome measures. This study provided evidence to support the choice of the final 9 items for the modified VAAI, but the new scale cannot be considered valid until the internal and external validity and test-retest reliability of its revised form are examined in new samples of patients. Furthermore, the construct validity estimates provided are for the 9-item VAAI, which were abstracted from the 81-item scale.

Conclusions

The original 81-item VAAI was shortened through EFA, item reduction, and internal consistency analyses to a 9-item questionnaire. The VAAI-9 appears to measure the construct of fear avoidance associated with physical activity and exercise, household and work responsibilities, and social activities quite satisfactorily. This may offer specific new insights into illness-related convictions and behaviors that limit patients’ engagement in therapy and restrict their return to important activities of daily living. The present study provided ample evidence for the internal consistency and construct validity of the final 9 items identified to be part of the revised VAAI. The inclusion of 9 items should not be burdensome for patients to complete in a clinical setting. However, before clinical use, additional research will be needed to validate the scale in its modified format, including examinations of its internal and external validity and test-retest reliability in patient cohorts from a broader range of clinical settings. Also, future research should focus on the ability of the 9-item VAAI to detect change over time and to evaluate the minimal clinically important difference of the scale. It will be important for clinicians to know what magnitude of change in VAAI score indicates clinical significance as well as the amount of change that is meaningful to the patient to improve clinical outcomes and quality of life for patients with vestibular disorders.