記住我
Context: There is a need for a clear cutoff to use noise sensitivity (NS) scale as a prediagnostic tool in research and clinical use. Aims: Noise sensitivity questionnaire (NoiSeQ) was adapted into Turkish language (Tr-NoiSeQ); validity and reliability of the Tr-NoiSeQ were assessed. The cutoff for the Tr-NoiSeQ was determined. Settings and design: A descriptive study within subject design. Materials and methods: A group of 402 participants aged between 18 and 52 years filled the Tr-NoiSeQ and the Turkish version of Weinstein noise sensitivity scale (Tr-WNSS) via internet. Statistical analysis used: Cronbach alpha (α) and the Spearman‒Brown coefficient were calculated for the reliability; Pearson correlation coefficient (r) and confirmatory factor analysis (CFA) were used for the validity. The receiver-operating characteristic (ROC) curve was used to calculate cutoff value of the Tr-NoiSeQ. Results: For the Tr-NoiSeQ, Cronbach α was 0.92, Spearman‒Brown coefficient was 0.93 (P < 0.05). A high significant correlation between the Tr-WNSS and the Tr-NoiSeQ was obtained (r = 0.76, P < 0.05). The fit indexes of the model were obtained as (χ2 = 1361.06, df = 546, P < 0.001), χ2 / df = 2.49, root mean square error of approximation = 0.06, comparative fit index = 0.83, goodness of fit index = 0.82. The ROC curve showed that the cutoff for the Tr-NoiSeQ was 53 with a 0.92 sensitivity and a 0.85 specificity (P ≤ 0.05). Conclusion: The Tr-NoiSeQ is a valid and reliable tool for use in both clinical practice and research. The Tr-NoiSeQ seems to be consistent with the Tr-WNSS and can be used as a screening tool to evaluate NS in audiology.
Keywords: Cutoff, hearing, noise, questionnaire, sensitivity and specificity
How to cite this article:Key Messages: Individual noise sensitivity can be quantified and classified by using a clear cutoff in scales. This quantification in terms of noise sensitivity helps to obtain more realistic and accurate results in clinical practice and researches.
Introduction
Noise can be defined as unwanted sound which may adversely affect the health and well-being of individuals. However, people’s attitudes and behavior toward noise can vary widely, indicating that people do not equally adapt to noise. This variability is usually explained under the term “noise sensitivity” (NS).[1],[2]
The NS has been defined as a state of sensitivity to noise or sound that emerges both as a stable personality trait covering an attitude toward noise and as an internal state that increases the individual’s degree of reactivity to noise.[1],[2],[3],[4],[5] NS is not a synonym of noise annoyance but a predictor of it. Because the reaction to noise in terms of NS may include not only annoyance but also apathy, frustration, depression, anger, exhaustion, agitation, withdrawal, and helplessness.[1],[2],[3],[4] It has been reported that estimated amount of variance in annoyance and dissatisfaction accounts for 10% to 26% of NS scores. In addition, individuals with high NS are more affected by noise, and they pay more attention to noise and perceive noise as more threatening.[2],[3],[4],[5]
Although there is no general consensus about the prevalence of NS, 17.1% of the participants reported NS in Widen and Erlandsson’s study with 1285 Swedish youth aged 13 to 19 years.[6]
The NS is an important issue studied by other disciplines related to audiology. As it can be genetic or familial, it is the subject of research in genetic studies related to the underlying causes.[7],[8] In addition, it may also have central origin. Kliuchko et al. examined the view that NS is associated with altered processing in the central auditory system by using magnetic resonance imaging in 80 participants with low and high NS scores in terms of self-administered NS questionnaire.[9] They reported that NS is related to enlarged primary auditory areas in the left hemisphere, enlarged bilateral hippocampus and temporal pole, increased volume of the right anterior insula, and emphasized that NS is related to the morphology of auditory-limbic brain areas.[9],[10]
In most reports attempting to explain NS in its more psychological and social aspects, it was associated with low scholastic skills, insecurity in social communication,[11] neuroticism, depressive symptoms, psychological distress, and being more sensitive to sensory stimuli.[12] Recent studies reported that NS can be a specific predictor of psychological health and the risk of psychological discomfort may increase depending on NS when exposed to road traffic noise[13] or the nonrestorative sleep can be predicted by NS especially in low-income families.[14] However, in mild traumatic brain injury study, it was reported that NS may have clinical use in marking vulnerability to persistent post concussive symptoms.[15]
From an audiological point of view, the relationship between hearing functions and NS has not been fully understood yet. Some reports on the relationship between NS and physiological correlates have shown that NS is not associated with auditory acuity in individuals with normal hearing.[2],[16],[17] Stansfeld did not find a significant difference in terms of NS in participants with normal and impaired hearing thresholds.[2] In addition, no significant correlation was found between NS and audiological findings such as pure tone hearing thresholds,[16],[17] loudness perception,[16] and distortion product otoacoustic emissions (DPOAE) amplitudes.[17] To evaluate the sensory component of NS in detail, Ellermeier et al. conducted a NS research with 61 participants aged between 19 and 37 years and used psychophysical measurement parameters including the mean absolute threshold, the threshold produced by the poorer ear alone, the difference threshold, the individual magnitude-estimation exponent, the slope parameter of the loudness category-scaling function, the mean unpleasantness rating of 10 sounds, and the slope of the function relating reaction time to SPL.[16] The 52-item psychometric questionnaire “Lärm‒Empfindlichkeits‒Fragebogen”[18] was used to assess NS. The results showed that only 15.2% of the variance in NS scores was explained by measurement parameters and there was a small but significant effect between NS and loudness scaling, and ratings of the unpleasantness of sounds. The researchers concluded that NS measured by a psychometric questionnaire was largely unrelated to the psychophysical variables of auditory functions. Likewise, Heinonen-Guzejev et al. did not find any significant differences in terms of NS between noise sensitive and non-noise sensitive subjects.[19] On the other hand, they reported that NS was associated with self-reported hearing impairment, especially among younger subjects, and higher NS scores were obtained as the self-rated hearing disability increased.[19]
Hyperacusis is the other important audiological issue whose relationship with NS has been investigated. NS is not a synonym of hyperacusis defined as an abnormally strong reaction to sounds perceived by the auditory system and manifests itself as a physical discomfort to low, medium, or high-intensity sounds.[20] As hyperacusis is related to the perceived loudness of the sound, NS is more about the processing of the noise, not the loudness of the sound. Neurophysiological studies suggest that hyperacusis occurs due to functional changes in the central auditory system and in the connections between the auditory system and other brain structures.[21] Similar to hyperacusis, recent studies showed that NS is related to the morphology of auditory-limbic brain areas referring that it may have a central origin.[9]
Summing up all these studies, individual variability in NS is an important issue that has been frequently studied in individuals with normal hearing[16] and hearing loss[22] as (i) an individual susceptibility in genetic research,[7],[8] (ii) a vulnerability factor in tinnitus,[6] and (iii) a personality trait in decreased sound tolerance including hyperacusis.[17] Moreover, hidden hearing loss due to synaptopathy at the level of afferent cochlear innervation of the auditory nerve cannot be detected by clinical, behavioral, or physiological measurements such as NS. However, it may cause perceptual complaints such as a decrease in the performance of speech understanding especially in a noisy environment.[23] Thus, NS should also be examined as a perceptual variable in hidden hearing loss.
To evaluate the NS in the field of audiology on an individual basis, there is a need for an assessment tool that serves to clearly distinguish between individual with and without NS. As NS is also reported in individuals with normal hearing and reflects a personality trait, the scales evaluating the NS have been standardized as psychometric tools measuring attitudes and behaviors. The Weinstein noise sensitivity scale (WNSS),[1] Turkish version (Tr-WNSS),[24] and the noise sensitivity questionnaire (NoiSeQ)[25] are two widely used psychometric tools to assess NS. To classify the quantitative distribution of NS scores on these scales, instead of a clear cutoff, individuals scoring in the lower 1/3 segment[1] or less than 25% quartile[25] have been classified as sensitive to noise, whereas individuals scoring in the upper 1/3 segment[1] or above 75%[25] were classified as non-noise sensitive. In such a classification, NS scores between the upper 1/3 and lower 1/3 or between 25% and 75% may cause uncertainty in identifying the presence of NS, consequently this uncertainty can also lead to potential confusions in the statistical analysis. In this regard, only Ellermeier et al., in which we previously discussed the relationship between audiological findings and NS, used a median score (med = 81) to distinguish between noise sensitive and non-noise sensitive individuals.[16]
In audiology and in its related fields, a psychometric NS scale can be used as a prediagnostic or screening tool to clearly discriminate the presence of NS by using a clear cutoff in terms of the complaints that the individual experiences perceptually and expresses subjectively. Therefore, in this study, it was aimed to determine a clear cutoff for the prediagnosis of NS by examining the relationship between the Tr-WNSS and the Tr-NoiSeQ. For this purpose, it was planned to adapt the NoiSeQ into Turkish language, to examine the validity and reliability of the Tr-NoiSeQ, and to determine a clear cutoff for the Tr-NoiSeQ.
Subjects and methodsSampling
In the present study, the minimum number of participants for the sample was planned as 10 times the number of items (1:10) of the Tr-NoiSeQ. Data were collected a total of 402 participants between the ages of 18 and 52 were reached.
Instruments
The noise sensitivity questionnaire
The NoiSeQ was developed by Schutte et al.[25] to measure global NS in the form of a 4-point Likert-type scale with 35 items in five areas of daily life: leisure, work, habitation, communication, and sleep. The reliability (relative and absolute G-coefficient) of this questionnaire is above 0.90, indicating that this questionnaire satisfies the precision level 1 “accurate measurement” according to ISO 10075-3. The internal consistency of the overall scale is as 0.90. According to the confirmatory factor analysis (CFA), all fit indices of four subscales (habituation, work, sleep, and communication) in the NoiSeQ indicate an acceptable match of the model.[26] The fit indexes of the model are as CMIN/df = 1.66, root mean square error of approximation (RMSEA) = 0.048, comparative fit index (CFI) = 0.860, goodness of fit index (GFI) = 0.86, adjusted goodness of fitness index (AGFI) = 0.835.
Linguistic validation for the Turkish version of the noise sensitivity questionnaire
In order for the NoiSeQ to be successfully adapted to Turkish (with the courtesy of Dr Schutte), back-translation and committee (controlled by experts in the field) methods were combined in accordance with cultural guidelines.[27] Initial translation from German to Turkish language was carried out by two bilingual translators who worked independently and by a German-educated expert audiologist. These three translated versions were compared and analyzed by the author and the other two audiologists to check for discrepancies between content and meaning of the original and the translated versions. After reaching a consensus, the final translation into Turkish was back translated German by a different bilingual independent translator. In a pilot study with 10 interviewers randomly selected within the university academic staff, the comprehension of sentences and words in the final translation form was checked. After the minor corrections, the instrument took its final form. The Tr-NoiSeQ was completed in a form of 4-point Likert-type scale with 35 items, from 1 (strongly agree) to 4 (strongly disagree) statements.
The Turkish version of Weinstein NS scale
The original version of the scale was developed by Weinstein at 1978.[1] Yildiz et al.[24] carried out the reliability and validity study for the Tr-WNSS, which consists of 21 items, with 210 participants (105 women and 105 men) between the ages of 18 and 55. Test‒retest reliability of the Tr-WNSS with 64 participants is as 0.92. The validity of the scale describes a two-factor structure, and the fit indexes of the model are as CMIN/df = 1.584, RMSEA = 0.054, CFI = 0.888, GFI = 0.892, AGFI = 0.866.
Administration of the questionnaires
The Tr-NoiSeQ and the Tr-WNSS forms were prepared as an electronic survey. The participants filled the questionnaires via the internet. By sharing the research link on social media including private groups such as youth and scientific groups, the opportunity to reach more participants was provided. It took about 15 minutes for a single participant to fill the questionnaires.
Statistical analysis
Statistical Package for the Social Sciences (SPSS) version 25.0 and SPSS AMOS version 25.0 (IBM Corp, Armonk, New York, USA) was used. Normality of distribution was evaluated by Kolmogorov‒Smirnov test and homogeneity of variances by Levene test. The reliability analyses of the Tr-NoiSeQ were examined with the Cronbach α and the Spearman–Brown coefficient. Pearson correlation coefficient was calculated to determine the validity coefficient. To assess the validity of the Tr-NoiSeQ, CFA was used. The fit indices of the model were tested with χ2/df , RMSEA, CFI, and GFI. The receiver-operating characteristic (ROC) curve analysis was used to calculate sensitivity, specificity, and optimal cutoff value of the Tr-NoiSeQ. All data were set of α = 0.05 significance level.
Ethical aspects
This study was approved by Başkent University Institutional Review Board and Ethics Committee (Project No: KA21/178). The purpose and the terms of participation of this research were declared to the all participants on the electronic questionnaire form. Informed consent was obtained from all participants by checking the "I voluntarily agree to participate in this study" option on the electronic questionnaire form.
ResultsDescriptive statistics
This study was carried out with 402 subjects (79% females, 21% males) between the ages of 18 and 52. Their mean age was 27.21 (standard deviation [SD] = 8.86) years. The Tr-NoiSeQ scores showed a normal distribution in terms of gender (for female Kolmogorov–Smirnov Z = 0.047, P = 0.089; for male Kolmogorov–Smirnov Z = 0.071, P = 0.200) and age (the Kolmogorov–Smirnov Z = 0.045, 0.059).
Psychometric analysis
The reliability analysis of the Tr-NoiSeQ was calculated by the internal consistency coefficient and the split half methods. The Cronbach α coefficient and the Spearman–Brown coefficient were found as 0.92 and 0.93, respectively (P < 0.05). In terms of subfactors in the Tr-NoiSeQ, the Cronbach α coefficient was as 0.61 for leisure, 0.71 for habitation, 0.78 for communication, 0.77 for work, and 0.87 for sleep (P < 0.05). This indicated that the Cronbach α values of all subfactors except leisure were obtained above limit value of 0.70.
To validate the Tr-NoiSeQ, the concurrent validity and the CFA were used. As given in the scatter plot presented in [Figure 1], the Pearson correlation coefficient between Tr-WNSS and Tr-NoiSeQ was found statistically significant (r = 0.76, P < 0.05), indicating a high correlation between Tr-WNSS and Tr-NoiSeQ. When examined the modification indexes of the Tr-NoiSeQ, the covariance was corrected between the 11th and 22nd items in the leisure subfactor, between the 4th and 33rd items in the habitation subfactor, between the 5th and 12th and between the 19th and 20th items in the communication subfactor. According to the CFA, the fit indexes of the model were obtained as (χ2 = 1361.06, df = 546, P < 0.001), χ2/df = 2.49, RMSEA = 0.06, CFI = 0.83, GFI = 0.82 [Figure 2].
Figure 1 The scatter plot between the Turkish version of Weinstein noise sensitivity scale (Tr-WNSS) scores and the Turkish noise sensitivity questionnaire (Tr-NoiSeQ) scores.
Figure 2 The confirmatory factor analysis of the Turkish noise sensitivity questionnaire with five subscales. C, communication; H, habitation; L, leisure; S, sleep; W, work.The cutoff for the Tr-NoiSeQ
To determine a cutoff for the Tr-NoiSeQ score in terms of NS, ROC curve analysis was used. Based on the Tr-WNSS scoring system, the highest 1/3 scores were identified as high NS and the lowest 1/3 scores were identified as low NS. After classifying the individuals participating in this study as with or without NS according to the Tr-WNSS scoring, a cutoff for the NS was determined depending on the scores of the Tr-NoiSeQ. As a result of the ROC curve, the cutoff value for the Tr-NoiSeQ was calculated as 53 with a 0.92 sensitivity and a 0.85 specificity (P ≤ 0.05). Area under the curve (AUC) was absolutely near the value of 1 (AUC = 0.94, P < 0.05), indicating a perfect and statistically significant value [Figure 3].
Figure 3 The receiver-operating characteristic (ROC) curve to determine a cut of point for noise sensitivity. Discussion The validity and reliability study of Tr-NoiSeQ was conducted to measure global NS as well as the NS for different everyday situations such as leisure, work, habitation, communication, and sleep.
When the internal consistency of the Tr-NoiSeQ was examined, Cronbach α coefficient for the overall questionnaire was 0.92 and the Spearman–Brown coefficient was 0.93. For the leisure, habitation, communication, work, and sleep subscales the Cronbach α values were found 0.61, 0.71, 0.78, 0.77, and 0.87, respectively (P < 0.05). These results showed that the items of the scale were analyzed consistently in terms of the individual’s self-assessment of NS and the Tr-NoiSeQ has a high reliability.
The reliability results of the Tr-NoiSeQ were found to be consistent with the reliability data of the original NoiSeQ. Similar to this study, in the original NoiSeQ, the internal consistency of the overall questionnaire (the Cronbach α = 0.90) and of all subscales except leisure subscale exceeded the lower limit (0.70).[25],[26] Schutte et al. reported that the reason why the leisure subscale does not meet this criterion was the different preferences between individuals in terms of leisure activities.[25] In addition, they stated that all items in the leisure subscale probably did not cover all possible activities, and that if the number of items increase, the reliability of this subscale could increase to 0.70.
According to the CFA, the Tr-NoiSeQ showed a multiple structure with more than five factors. However, to fit the factor structure as in the original form, all items in the Tr-NoiSeQ were grouped into five subscales. The first two values (χ2 = 1361.06, df = 546, P < 0.001, χ2/df = 2.49, RMSEA = 0.06) of the model fit index were within acceptable limits and the last two values (CFI = 0.83, GFI = 0.82, P < 0.05) were obtained close to acceptable limits. However, there was a high correlation between the Tr-WNSS and the Tr-NoiSeQ (r = 0.76, P < 0.05). As a result, the Tr-NoiSeQ has been validated with its five-factor structure.
Ellermeier et al. concluded that in old reports evaluating NS and auditory function, the lack of using psychophysical measure and insufficient number of participants may lead to an underestimation of the relationship between auditory functions and NS.[16] In addition to this conclusion, we would like to emphasize that there was uncertainty in terms of people having NS according to scale scores. As NS has been considered as a personality trait in psychological and sociological studies and has been examined on the basis of attitude and behavior, there was no fixed cutoff in terms of NS scale scores and scoring was made between upper and lower values. However, in audiology, individuals with and without NS should be differentiated to make a prediagnosis. Therefore, with this study, a cutoff was determined for the first time on a scale evaluating NS. The statistical cutoff for the Tr-NoiSeQ scale scores was 53; individuals with NS scored 53 and above, whereas individuals without NS scored less than 53. As NS in audiology cannot be evaluated with an objective test method, it can be stated that Tr-NoiSeQ can be used as a reliable screening tool to distinguish the individuals with and without NS. Consistent with the cutoff, Ellermeier et al. used a median value and reported that participants with high NS also had higher hearing thresholds, although it was not statistically significant in their study.[16] However, in this study, only 1 kHz hearing threshold was obtained with psychoacoustic methods, and the relationship between NS and high frequency hearing thresholds was not investigated.[16]
In the present study, demographic factors such as age and gender were not evaluated as the main purpose is to identify individuals with and without NS. However, on reports examining the relationship between the NS and demographic factors, there has been no clear consensus in terms of age and gender.[6],[28] For example, Widen and Erlandsson examined the prevalence of tinnitus and NS among young people in Sweden.[6] They statistically evaluated the NS in a group consisting of 3254 participants in the 13 to 19 age group. They reported that 216 participants (17.1% of the total group studied) had NS, the prevalence rate of NS found higher among upper secondary school students (19.7%) compared to younger students (14%), and this difference was found statistically significant (F = 4.93; df =░ P < 0.01; N = 1263). Moreover, Dzhambov and Dimitrova conducted an adaptation study in which the Bulgarian version of the NS short form was examined for validity and reliability.[28] In their study, the NS scores of 115 participants with a mean age of 45.42, SD = 15.8 years were evaluated. Interestingly, they concluded that the scale scores were higher in younger people (r(69) = –0.363, P = 0.002, 95% confidence interval: −0.649 to −0.077).[28] In terms of gender, contrary to older reports showing no gender effect on NS,[29],[30] newer studies indicate that females have higher NS than males.[19] Also, Heinonen-Guzejev et al. reported that NS was associated with self-reported hearing disability among women, and that the audiometric data of elderly female subjects demonstrated somewhat better hearing levels among noise sensitive women. It was also emphasized that the relationship was primarily reported among young subjects (50 years and under). These variations in the relationship between demographic factors and NS may be due to the fact that a cutoff point for NS was not determined in previous studies.
ConclusionIt has been shown that the Tr-NoiSeQ is reliable and valid tool to evaluate the subjective NS. However, it proved to be somewhat problematic with respect to its leisure-based items. To increase the reliability of this subscale, it is necessary to increase the items. However, the Tr-NoiSeQ seems to be consistent with the Tr-WNSS and it can be used as a screening tool with a specific cutoff to evaluate NS both in clinical practice and research. However, the relationship between audiological findings and NS may have been underestimated in reports that do not use a specific cutoff for scoring and have insufficient number of participants. Further research is needed to identify individuals with or without NS according to a specific cutoff, to reveal the audiological profiles of these individuals, and to interpret them in terms of demographic factors such as age and gender.
Acknowledgment
The authors express their deepest appreciation to all those who participated in this study.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
References
Correspondence Address:
Asuman Alniacik
Faculty of Health Sciences, Department of Audiology, Başkent University, Bağlica Campus, Eskisehir Road, 18.km Fatih Sultan Mehmet Street, Ankara, 06810
Turkey
Source of Support: None, Conflict of Interest: None
CheckDOI: 10.4103/nah.nah_46_21
留言 (0)