記住我
Background: One of the most common hearing disorders is hyperacusis, characterized by decreased sound tolerance or noise sensitivity. Objective: The present paper aimed to evaluate the validity of the Persian version of the hyperacusis questionnaire. Methods: The sample was comprised of 434 students evaluated at the Kashan University of Medical Science, Isfahan, between July 2019 and February 2020. After translation and retranslation of questionnaire, the content validity was assessed by 15 occupational health experts using content validity index (CVI) and ratio (CVR). Validity and reliability of the scale were examined by using confirmatory factor analysis, the Cronbach alpha, composite reliability (CR), and the average variance extracted (AVE) tests. Results: In the exploratory analysis of the Persian version of Khalfa hyperacusis questionnaire, 14 items were created based on three subscales) attentional, social, and emotional dimensions) which explained 46.5% of the total variances. Content validity of 14 questions was approved with CVR > 0.49 and CVI > 0.7. The Cronbach alpha, AVE, and CR indexes were calculated 0.811, 0.761, and 0.901, respectively, which showed that reliability of the scale was adequate. Conclusions: It was concluded that this Persian version of the hyperacusis questionnaire has relatively acceptable validity and reliability in Iran. Given that the tool has a small number of questions, it is easily used in different studies.
Keywords: Hyperacusis, psychometric, questionnaire, translation, validation
How to cite this article:
One of the most common hearing disorders is hyperacusis, characterized by decreased sound tolerance or noise sensitivity,[1],[2] occurring with both low-pitched and general tones.[3] It is often associated with hearing loss and tinnitus.[3],[4],[5] People with hyperacusis find sounds considered comfortable by others to be loud and unbearable.[2],[6] Intolerance to sound causes disruption in social activities, work, and even recreation.[7] In studies, the prevalence of hyperacusis has been reported between 2% and 15%.[8],[9] The disorder can affect all age groups, even children. The prevalence of reported hyperacusis is directly related to aging on the social dimension.[9],[10] The causes of hyperacusis are unknown.[11] The most common causes of hyperacusis include cochlear injury, head trauma, hearing damage due to medications or chemicals, hearing loss, old age, surgery, chronic ear infections and autoimmune disorders, and other physiological lesions. However, it has been shown that clinical hyperacusis can cause intolerance to sound even in the normal hearing threshold,[12],[13],[14] which can be caused by various injuries of the peripheral auditory system, central nervous system, and infectious diseases.[7],[15],[16],[17] Various methods have[17] been proposed for the treatment of hyperacusis, including sound therapy and cognitive behavioral therapy. For the initial diagnosis of the disease, an audiometric test is performed. In this test, in addition to measuring the auditory threshold at each frequency, the lowest disturbing level of sound is measured. Psychometric and self-report tools are also used to measure hyperacusis.[17],[18],[19] One of the most reliable tools for measuring hyperacusis is hyperacusis questionnaire by Khalfa et al.[14] This tool has not been translated into Persian and its psychometric properties have not been evaluated. Given the importance of hyperacusis and its high prevalence, and due to the fact that the use of this tool does not require special devices and facilities, this study aimed to evaluate the validity and reliability of Khalfa hyperacusis tool.
MethodsStudy design
The present descriptive study examined the validation of the Persian version of the Khalfa hyperacusis questionnaire (KHQ). This study was approved by Kashan University of Medical Sciences (Grant number = 97028) and approved by the Ethics Committee (IR.KAUMS.NUHEPM.REC.1397.028).
Sample size
The study was performed on 434 students of Kashan University of Medical Sciences. Based on the study by Anderson et al.[20] in 2002, and the prevalence of 17.5% hearing sensitivity among high school students,[20] 95% confidence, and 5% accuracy, the sample size was calculated 217 people considering the stratified sampling coefficient, this number reached 434 people. According to MacCallum et al.,[21] the minimum sample size for exploratory studies should be between 100 and 200 people. According to Comrey and Lee,[22] in studies that examine confirmatory factor validity, 200 people are relatively good and 300 people are good.
Instrument
The instrument used in this study was a hearing sensitivity questionnaire that Khalfa et al.[14] first designed in 2002 and has been validated in many studies and translated in Arabic,[2] Italian,[23] and Japanese,[24] and its psychometric properties have been evaluated. This tool measures a person’s hyperacusis in three fields of attentional, social, and emotional dimensions. Attentional and emotional dimension fields have four questions and social dimension field has six questions that have a cutoff point of 28. The answers of the items were scored based on a 4-point Likert scale, including no (1), low (2), high (3), and very high level (4).
Linguistic validation
Translation and retranslation of the tool was performed according to the protocol proposed by the World Health Organization (WHO). Two native translators independently and separately translated the English version of the tool into Persian. Both translators had a PhD in English language and one of them was familiar with vocabularies of occupational health. At the end of this stage, two Persian versions of the tool were obtained from the two translators. Then, the Persian version was retranslated into English, and the translators, along with an occupational health expert, compared the version to the original text ([Figure 1] and [Figure 2]).
Figure 2 Second order factor analysis of the Persian version of the KHQ.Content validity
To assess the content validity, 15 occupational health experts with at least a master’s degree were used to provide their opinions for each of the questionnaire items. To assess the necessity of each question, the experts’ opinions were considered based on a 3-point Likert scale, including “necessary,” “useful but unnecessary,” and “useless,” and based on their answers, the content validity ratio (CVR) was calculated:
In which, Ne is the number of experts who chose the “necessary” option and N is the total number of experts evaluating the questionnaire. The confirmation criterion of each question was considered CVR = 0.49 based on Lawshe table.[25] Moreover, three criteria, including simplicity, relevancy, and clarity of each item were scored based on a 4-point Likert scale from low (1) to very high (4) by the experts. The content validity index (CVI) for each of the questions was also calculated.[26]
The numerator of the fraction is the number of experts who gave each item a score of four and three, and the denominator of the fraction shows the total number of experts. According to the method proposed by Waltz and Bausell,[27] a CVI score >0.79 is appropriate, a score between 0.7 and 0.79 needs to be revised, and a score <0.7 is unacceptable and should be deleted. After this stage, by creating a focus group, including occupational health experts (n = 3 and statistics experts (n = 1), all questions with CVI < 0.7 or CVR < 0.49 were reviewed and modified.[28]
Exploratory factor analysis
At this stage, the Kaiser–Meyer–Olkin (KMO) statistic and the Bartlett test were used to evaluate the adequacy of the samples and the correlation of the data. A KMO value > 0.7 is an estimate of the adequacy of the sample size. However, the KMO and Bartlett test both indicate that the set of variables are at least adequately related for factor analysis.[29]
Exploratory factor analysis (EFA) is the most common method for determining the validity of questionnaires and categorizing questions to different scales. Therefore, the questionnaire developed in the present study was reviewed using EFA with principal component analysis (PCA) and varimax rotation.[30],[31]
Confirmatory factor analysis
Confirmatory factor analysis (CFA) was used by likelihood maximum estimation using AMOS (New York, USA) statistical software version 22. The CFA seeks to answer the question of to what extent the assumed or predicted relationships between the variables are consistent with the relationships in the observed actual data. If two correlation matrices (assumed or proposed correlation matrices and correlation matrices obtained from actual data) are identical, the proposed model will be a valid explanation for the assumed relationships. To evaluate the goodness-of-fit of the three-factor model of the study, chi-square (χ2), χ2.df, root mean square error of approximation (RMSEA), comparative fit index (CFI), goodness-of-fit index(GFI), relative fit index, incremental fit index (IFI), and normed fit index (NFI) were used. Based on the mentioned indicators, the numerical value <3 for χ2.df, 0.06 for the RMSEA, and value >0.9 for other indicators indicate good model fit. In the present study, in order to improve the model fit indices, according to the software suggestions, the model modification indices were used to modify the model.[32],[33]
Reliability
The reliability of a tool means that the results obtained from it are consistent in repeated measurements. The most common test for measuring reliability is the internal consistency (Cronbach alpha), which is based on the correlation between items in a dimension or dimensions in an index. The closer the alpha level to 1, the more reliable the questionnaire. The acceptable value for this correlation was considered 0.7.[20],[34]
Composite reliability (CR) or stability of new factors in structural equation modeling (SEM) is a better criterion than Cronbach alpha. It is due to the fact that in calculating Cronbach alpha coefficient for each construct, all indices are entered with equal importance in the calculations. However, in CR, indices with a higher load factor are more important. CR is an alternative to Cronbach alpha coefficient in SEM. In the present study, if CR values are >0.7, the average variance extracted (AVE) is >0.5, and if CR values are greater than AVE (CR > AVE), CR was considered acceptable.[35],[36]
Convergent validity assessment of the construct was measured by Fornell and Larcker[35] approach using the AVE and the maximum shared squared variance (MSV). In order to establish convergent validity, AVE must be >0.5. This indicates that the construct explains ≥50% of the variance of its markers, and to confirm divergent validity, MSV must be less than AVE.[35],[36]
In this study, 434 students participated, 157 of whom (35.3%) were male and the rest were female. All the students were between the age range of 18 and 36 years. The mean age of the students was 21.66 years and the standard deviation was 2.36. Most of the students were undergraduate (279) and professional doctorate students (132) and the least number of them were medical specialist (2) and postgraduate[14] students [Table 1].
After translation and retranslation, the questionnaire was approved by English language and occupational health experts in terms of face validity. In terms of content validity, in the first stage, 14 questions were approved with CVR > 0.49 and CVI > 0.7. The CVR value varied from 0.73 to 1 [Table 2].
Table 2 Questions of Persian version of hyperacusis questionnaire, face and content validity indices, factor loading, explained variance, and reliability indexConstruct validity
Exploratory factor analysis: The KMO statistic was 0.839 and the bartletts test (BT) test was statistically significant (P < 0.001), indicating the adequacy of the data for exploratory analysis and also showed that the data correlation matrix in the population was not zero; therefore, it is possible to act to find factor. Based on Kaiser criteria, values >1 were considered for factor extraction. Based on the results of the PCA, the whole questionnaire was based on three subscales, explaining 46.524% of the total variance [Table 1]. These three subscales were named attentional dimension, social dimension, and emotional dimension based on the original version. The PCA also determined the items related to each of the subscales based on the varimax rotation results and according to the maximum participation weight. Four questions were in factor 1, six questions in factor 2, and four questions were in factor 3. Therefore, at the end of EFA, there were a total of three factors and 14 questions in the set of questions and no question was omitted [Table 2].
Confirmatory factor analysis: [Table 3] shows that the χ2 value was 125.334, which was significant. Given that the most important fit statistic is the χ2 statistic, this statistic measured the difference between the observed and estimated matrices. This statistic is very sensitive to the sample size, so its value is divided by the degree of freedom. If the result is <2, it is appropriate. [Table 3] reveals that this value was <2. Another GFI is the RMSEA, which was 0.04 in the present study and was acceptable, since it was <0.05, indicating the approval of the research model. Other indices, including Tucker–Lewis index, NFI, IFI, and CFI, all confirmed the acceptability of the first and second order model.
Table 3 Fit indices of the first and second order model of the Persian version of the hyperacusis questionnaireReliability: Cronbach alpha coefficient was used in the construct reliability analysis. The calculated coefficient was 0.811 in the Persian version of the KHQ. [Figure 1]–[Figure 2] reveals the first and second order confirmatory factor analysis of the 3 factor structure of the questionnaire. it showed that all indicators loaded highly (>0.3) and significant (P<0.001) suggesting that both models were convergent validity. Analyses indicated that the AVE was equal to 0.761 and the CR index of the questionnaire was equal to 0.901, which was greater than the AVE. Therefore, the convergent validity of the construct was confirmed. The MSV was smaller than the AVE, confirming the divergent validity of the construct.
DiscussionThis study aimed to translate the KHQ[14] into Persian and examine its psychometric properties.
In the exploratory analysis of the Persian version of KHQ, 14 items were created based on three subscales, which explained 46.5% of the total variances. Each of the subscales had 29.5%, 9.7%, and 7.2% variance explanation, respectively. Explanatory power of 46.5% is relatively low for a valid tool. Khalfa et al.[14] used EFA technique to determine the psychometric properties of the hyperacusis questionnaire. Their results showed that 14 items in three subscales of attentional dimension, social dimension, and emotional dimension had values >1, which explained 48.4% of the total variance. The score of each subscale was 27%, 13.8%, and 7.6%, respectively.[14] The Persian version of KHQ was without any change in the number of questions resulting from translating questions from English to Persian. Low explanatory power of variance was also seen in the study of Khalfa et al.[14]
Several questions need to be answered when examining the psychometric properties of the translated questionnaire. Internal validity, which is often included in studies, is mainly assessed using Cronbach alpha. In the present study, Cronbach alpha coefficient was appropriate for the whole tool (0.811) and in the study of Erinc et al.,[37] the coefficient of the Turkish version of the Khalfa questionnaire was 0.81. The Italian version of the questionnaire also reported a Cronbach alpha coefficient of 0.89.[23] The Japanese version also reported an internal consistency coefficient of 0.92.[24] Nunnally and Bernstein[38] stated that in case of using a new questionnaire, its alpha coefficient should be at least 0.7. The Cronbach alpha similarity of various studies indicates the intercultural validity of the adapted scale, which adequately reflects the effectiveness of the original English version.
Cronbach alpha should be separately calculated for each subscale, since each subscale should be able to express the main concept related to the subject.[39] In studying the reliability of attentional dimension, social dimension, and emotional dimension subscales, Cronbach alpha coefficients were 0.667, 0.66, and 0.6, respectively. These values in the original version were calculated to be 0.66, 0.68, and 0.67, respectively.[14] The reliability of the questionnaire in the subscales seems to be relatively low. Reliability can be increased through conventional methods, such as increasing the number of questions or items, changing questions or items, and making them more homogeneous. Despite the language differences, the two versions had an acceptable adaptation. To use this tool in clinical situations, it is required to increase its reliability.
In this study, in order to evaluate the validity of the three-factor structure of KHQ, CFA based on the SEM was used. The results of first-order CFA showed that the high load factor of the items were statistically significant. Moreover, the fit indices of the measurement model (χ2/df) and RMSEA were equal to 1.74 and 0.04, respectively, indicating the overall adequacy of the measurement model. These results indicated that the three-factor structure of the questionnaire had the best overall fit, even without modification. This finding is consistent with the results of previous studies. In the study by Erinc et al.[37] in the Turkish version, the values of χ2/df, GFI, and RMSEA were equal to 3.89, 0.91, and 0.068.
There were some limitations in this study. First, the participants were limited to a group of students who were selected by the convenience sampling method, which limits the generalization of the KHQ tool to other groups, including industry workers. Therefore, it is recommended to perform the factor structure of the questionnaire in other groups in future studies. Another limitation is the lack of use of parallel tools or clinical assessment, such as audiometry in psychometric analysis, which physicians and researchers may be cautious about clinical evaluation. It is recommended to use other methods in future studies to increase the validity of the research and determine the appropriate cutoff point and sensitivity analysis.
This study showed that the Persian version of the hyperacusis questionnaire has relatively acceptable validity and reliability in Iran. Given that the tool has a small number of questions, it is easily used in different studies. On the other hand, this tool has been used in different parts of the world and, therefore, provides the ability to compare the results of domestic studies with other countries. Because this study involved the hyperacusis questionnaire between the students of 18 and 36 years of age range, it is recommended that validation be done in future studies in other age groups.
ConclusionThe Persian version of the KHQ has the appropriate construct validity compared to the original version without changing the number of questions and subscales. Due to the small number of questions, its usability is high; however, to increase the construct validity, an expanded version can be suggested.
Financial support and sponsorship
Authors would like to thank Vice Chancellor of Research and Technology, Kashan University of Medical Sciences for providing financial support to conduct this work (Approval code: 97028).
Conflicts of interest
There are no conflicts of interest.
References
Correspondence Address:
Hossein Akbari
Associate Professor of Biostatistics, Trauma Research Center, Kashan University of Medical Sciences, Kashan
Iran
Source of Support: None, Conflict of Interest: None
CheckDOI: 10.4103/nah.nah_16_22
留言 (0)