SPEECH PERCEPTION TEST HINT BRAZIL IN GROUPS OF SUBJECTS EXPOSED AND NOT EXPOSED TO OCCUPATIONAL NOISE

(1) Speech Therapist; PhD in Public Health, Faculty of Medical Sciences, University of Campinas – UNICAMP, Campinas, SP, Brazil; Master in Public Health Epidemiology at the University of Campinas, UNICAMP. (2) Speech Therapist; Professor of Audiology at the University of Campinas, SP, Brazil; PhD in the Faculty of Medicine, University of São Paulo – USP. (3) Otorhinolaryngologist; Collaborating Professor of Otorhinolaryngology, Head and Neck Surgery, Faculty of Medical Sciences, University of Campinas – UNICAMP, Campinas, SP, Brazil; PhD in Public Health from the University of Campinas – UNICAMP.


INTRODUCTION
Speech perception investigation has become a widely discussed subject lately and the necessity of performing tests that simulates a real listening situation became fundamental in order to evaluate the impairment challenges of each subject 1 .workers 7 , beyond the fact that is part of great discussions regarding public health in Brazil 8 .Speech recognition of sentences in silence and noise in workers with NIHL may appear in lower degrees of loss and increase significantly in higher degrees 9 .
Speech perception tests with competitive noise in low redundancy situations that contemplate a wide analysis of hearing disabilities and amplify logo-audiometric tests findings, are not frequently used in the country, especially in subjects exposed to occupational noise 10,11 .
The Hearing in Noise Test (HINT) appears as a strategy to improve the Speech Recognition Test (SRT) metering.Developed in 1994 12 , it has been utilized in different languages and population groups.In Brazil, recently standardized in normal hearing groups 13 , is based in an adaptive test technique, where the sentence presentation level increases or decreases accordingly to a spectral noise in own speech 14 .Speech perception test are realized with monosyllables or polysyllables routinely.It's really important the use of phonetic context sentences with HINT, word familiarity, intonation variation and influence level intelligibility in noise.In many countries HINT is used in workers screening, especially where hearing plays a vital or important role in the job 15,16 .
The purpose of this study was to research the differences in speech to noise ratio in subjects exposed and not exposed to occupational noise with and without hearing loss by using the HINT, version in Portuguese from Brazil.

METHOD
After the approval of the local Research Ethic Committee, 206 subjects were analyzed at the Occupational Otorhinolaringology Ambulatory and were divided in 3 groups: G1 -56 subjects with normal hearing non-exposed to occupational noise; G2 -70 subjects with normal hearing exposed to occupational noise; G3 -80 subjects with sensorineural hearing loss exposed to occupational noise.
G3 was initially divided in two subgroups: subjects exposed to suggestive and non-suggestive occupational noise NIHL in order to verify the homogeneity between these subgroups.Mann-Whitney statistical analyzes demonstrate homogeneity in every variable.Thus, G3 was classified as a hearing loss subject group.Below are the results.The exclusion criteria to every group were: neurological and/or verbal fluency changes, middle ear impairment, cerumen stopper or sentences hearing difficulty.To G1 and G2: adults or elderly with hearing loss and complains on speech perception in noise.To G3: adults or elderly with mixed or conductive and unilateral hearing loss.
An external auditory canal inspection was done in order to guarantee any change in the middle ear.Participants were submitted to a basic hearing anamnesis, threshold tonal audiometry, followed by the application of the speech tests using headphones.The results processed by the system were stored.
HINT utilizes a microprocessor HTD (Hearing Test Device) version 7.2 Audiometric System, produced by Bio-Logic, developed at House Ear Institute, Los Angeles, USA, in 1994 12 .
HINT contains 12 lists of 20 sentences each, digitally recorded which can be presented in silence and noise.Those sentences are standardized accordingly to the language, difficulty, intelligibility and phonetic distribution.
The test application varies from 2 minutes for the 12 sentences list and 3-4 minutes for the 20 sentences list.Sentences were presented by a male speaker in quiet and in noise fixed at 65 dB (A), accordingly to the established standards.
The equipment presents the recorded sentences with a male professional voice and a competitive noise created from the same voice spectrum.It also conducts the whole test procedure, including the tonal audiometry.
Speech and noise were presented with headphones in four conditions: Quiet (Q), noise front (NF), noise right (NR) and noise left (NL).The processor also calculates the noise composite (NC) by the weighted average of the three noise results: Each subject sits inside an audiometric booth in order to perform the test.The previously recorded signal and noise stimuli are played to the headphones by the equipment programmed to simulate the stimuli localization.
HINT selects randomly 1 in 12 sentence lists and its presentation follows the up-down strategy, which allows the determination of the speech recognition threshold (SRT) to the signal/noise ratio, established at 50% 14 .The first four sentences are presented with a rate of 4 to 4 dB which estimates the subject threshold.From the fifth sentence on, the rate is changed to 2 dB and the definite threshold is determined after the presentation of the 20 sentences to each condition of the test.
The participants must repeat the listened sentence completely in order to be accepted and the examiner must compute (yes or no) each presented sentence so that the program calculates the signal/ noise ratio in the system.
Subjects were tested with headphones to compare the results of the normal hearing groups in Brazilian Portuguese and with other works involving the current material.
The results were calculated by the signal/noise metering.It's important to point that the more negative it is, greater is the difficulty of the test, due to the speech is lower than the presented noise.
A descriptive analysis with the presentation of frequency distribution tables to categorical variables and position and dispersion measurements to numerical variables was presented.In order to compare the proportion, we used the Qui-Quadrado test or Fischer exact test when necessary.To compare continuous or ordinal measurements between two groups we used Mann-Whitney test and between three groups, the Kruskall-Wallis test.To verify the linear association between the measurements, we used the Sperman Correlation Coefficient.This coefficient varies from -1 to 1 and the values close to the extremes indicate negative or positive correlation, while the values close to zero indicate no correlation.The significance level adopted to the statistical analysis was 5% 18 .

RESULTS
The relation between HINT application conditions and noise exposure time in G2 showed significant difference in Spearman linear coefficient in S (c= 0,26568; p= 0,0262) e NC (c=0,29413; p= 0,0135).In G3, the statistical results were not significant.
Table 1 shows the distribution of 500, 1000, 2000, 3000, 4000 and 6000 Hz average thresholds in both ears of each group.
The average of 500, 1000, 2000, 3000, 4000 and 6000 Hz frequencies from both ears were compared to verify the threshold differences between the groups using Spearman linear correlation coefficient.A correlation between HINT performance and the averages described was found, showing that the worst is the average, worst is the HINT performance.In G1, correlation for the Q condition was found; in G2, correlation for the 500, 1000 and 2000 Hz averages in the conditions Q, NR, NL and NC, and for the 3000, 4000 and 6000 Hz averages in the condition NL were found.Such results are shown in the Table 2 Table 1 -Hearing threshold averages by frequency average and their respective standard deviation to the three groups Hearing threshold arithmetic averages (AA) and their standard deviation (SD) in groups: G1 -normal hearing group not exposed to noise; G2-normal hearing group exposed to noise and G3 -hearing loss subject group; right ear (RE) and left ear (LE) A statistical difference (p=0,0001) was found only in the NC condition between G1 and G2 by performing a descriptive analysis and comparisons between the groups on HINT test performance with headphones using the Krukal-Wallis method..

Group
Figure 2 shows the HINT Brazil values on each condition per group.Despite the answers beneath the normal thresholds for every group in the logoaudiometric tests, an inferior performance was found in G3 when compared to G1 and G2.This results shows that hearing loss subjects present an inferior relation signal/noise to the HINT Brazil test.
Comparative tests between G1 and G3, and also between G2 and G3 showed differences in every test conditions (p<0,0001) by the Kruskal-Wallis method.
The average values in normal hearing subjects in Brazilian studies considering the four conditions in HINT Brazil performance are described in the      It's important to remember that SSRT and NSRT use the same elements as HINT: fixed noise at 65 dB(A) and sentences with speech spectrum noise, but the difference is in the form they are applied, because HINT allows the sentences presentation in software, different test conditions, computes the signal/noise ratio fast and doesn't need to be applied by CD and/or audiometer.The authors also suggest routine tests application to measure the subject group speech perception.
The findings are similar to those found in a normal hearing and hearing loss subjects research regarding the worse speech perception performance from the hearing loss subjects when compared to normal hearing and these results are not detected in logo-audiometric tests applied in the clinic 27 .
We observed correlation between HINT performance in Q condition and the audiometric thresholds averages, being in agreement with findings that indicate that both tests can predict the intelligibility.However, HINT offers more information as it evaluates the signal/noise ratio with speech spectrum noise sentence stimuli and enables a substantially precise measure in different population groups in different languages 28 .

CONCLUSION
Our study found that there is difference regarding HINT Brazil test performance in normal hearing subjects exposed and not exposed to occupational noise in the Composite Noise condition.The worst HINT performance in every application condition was in the hearing loss group when compared to normal hearing subjects with and without noise exposure.
HINT using shows to be an efficient method to verify speech perception difficulties in noise in the hearing loss group.Nevertheless, further research are necessary in order to disseminate and apply HINT Brazil method in the clinic practice.

Figure 2 -
Figure 2 -HINT values to the conditions: Quiet (Q), Noise Front (NF), Noise Right (NR), Noise Left (NL) and Noise Composite (NC) for the three groups.