Prevalence of hearing loss in adolescents and young adults as a result of social noise exposure : meta-analysis

The exposure to noise in the teens and young adults leisure has drawn attention, given the impact of hearing loss in this population. This study had the purpose to estimate of the prevalence of hearing loss in this population resulting from exposure to social noise. Seventeen articles were identified for analysis that met the selection criteria on which it was observed information as design, age, location, evaluation mode, and the prevalence of hearing loss. The prevalence in the self-reported studies was less than 2% while those carrying audiometric ranged from 11.5 to 15.8%. It is concluded considerable heterogeneity between the prevalence of self-reported hearing loss and the measured by audiometric tests in the studied population.


INTRODUCTION
One of the biggest public health problems currently is the noise pollution and its effects on health, including hearing which, for this reason, has been broadly investigated 1,2 .According to the World Health Organization (WHO), over 5% of the world population -360 million people -have disabling hearing loss (328 million adults and 32 million children).And it is estimated that most of these people live in developing or underdeveloped 3 countries.One of the most common causes of hearing loss are German measles infection during pregnancy, ear infections and prolonged exposure to loud noise that stands out as a leading cause of irreversible hearing loss 4,5 .
There is an increase in hearing loss in children and adolescents related to recreational noise exposure 1 .Many young people willingly expose themselves to high levels of noise intensity because of their habit of listening to music in bars, nightclubs, parties, syndromes, auditory neuropathy and / or middle ear pathologies.

Collected data
The reading of selected articles was carried out by two researchers using a checklist for studies assessment and a standard form which included the collection of the following information: authors, year of publication, country and population studied, design, sample size and description, participants' age range, hearing assessment mode, and prevalence of hearing loss in the age group of interest.

Study score analysis and assessment
The extracted data were compared by taking into consideration country of origin, type of study, sample size, participants' age group, type of auditory assessment and main observed results.
For the study scores assessment, the factors that could be potential sources of error were taken into consideration, such as sample size and method for hearing loss assessment (self-report or measured by instrument).
Studies received a score, considering the following criteria established by the authors: complete audiometric evaluation, incomplete (the ones that had only air conduction) or self-report; study population (population or specific group); age (age range or a specific age); protocol (WHO or others); and sample size (small, medium or large).From the score of each study, a score was established by quartiles, upper (above the third quartile), moderate (between the first and the third) and lower (below the first quartile).

Statistical analysis
A meta-analysis was carried out for those studies that showed the prevalence of hearing loss in certain age range.The heterogeneity of the studies was assessed by the inconsistency (I²) which depicts the percentage variation attributed to them beyond chance.The following cut-off points were considered: 25% low; 50% moderate and 75% high heterogeneity, with p-value <0.05.Due to the high variability in the data, the random effect model was adopted.Initially, the prevalence was transformed into logit function to calculate its average.Then, they were weighed by the inverse of its variance.We used the Forest graphic for the presentation of the meta-analysis results and comparison of studies.Analyses were performed on the "meta" package from R software.

Search strategy and selection criteria
Literature systematic review, with the descriptors DeCS (Health Sciences Descriptors -BIREME) and MeSH (Medical Subject Headings, PubMed) was used without restriction: prevalence(s) AND (hearing disorder(s) OR dysacusis OR hypoac$ OR hearing impairment OR bilateral deafness OR hearing loss, complete OR acquired deafness OR hearing loss, extreme) AND (adolescent(s) OR teen(s) OR teenager(s) OR teenager youth OR youths adolescence OR adult(s), young).Due to the differences of search mechanisms and uniterms recognized by each database, it was necessary to adjust the use of descriptors, for instance, including prevalence in the search on Web of Science.The search of articles was carried out in electronic databases -MEDLINE via PubMed, LILACS, WEB OF SCIENCE, SCOPUS and SCIELO.
The inclusion criteria regarded articles published between January 2000 and July 2013, with available abstract, with free or paid internet access.The search was conducted without any language restriction.Although later, only studies published in Portuguese, English and Spanish, which assessed the prevalence of hearing loss in adolescents and young adults, have been included in the phase of selection.The age classification followed the criteria of the World Health Organization which covers the period of adolescence in the age range of 10 to 19, while youth corresponds to the age group between 15 and 24 years old.The term "young adult" is also used referring to someone who is between 20 and 25 years old 13 .

Eligibility for studies
A screening of the localized studies was held, following the methodological steps proposed by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses -PRISMA 14 , in order to select the texts that reported the prevalence of hearing loss due to noise exposure in adolescents and young adults.Two authors (A1 and A2) proceeded independently, and in a standardized manner, the selection of items identified in the databases, being eligible those which, by reading the title and / or abstract, contained data on the prevalence of hearing loss in the age group of interest (10 to 25).The divergences found were resolved by consensus, and the selection of articles to be read in full was established.
The exclusion criterion of the reading protocol selected articles, which had data of hearing loss associated with pre-existing conditions, genetic changes, cochlear implants, multiple disabilities, 42,651 (forty-two thousand six hundred and fiftyone) have been reported with some level of hearing loss.Almost all the articles beheld men and women, except the ones about military service tracking which reported only results of young men.Three studies showed a higher prevalence of hearing loss in men.
Concerning the design of the study, nine population surveys have been identified, a survey for scholars, another one for health service users.Five papers of tracking were selected, three on the assessment for military service, one on evaluation for job admission to enter an industrial workplace and the last one on university music students (Table 1).

LITERATURE REVIEW
The electronic search in databases resulted in the identification of 438 articles.Twelve (12) studies were excluded due to duplicity, remaining 426 articles.Out of these, 354 were excluded because they did not have the criteria for inclusion after assessment of titles and abstracts.Among the 72 articles selected for full reading, 17 met inclusion criteria (Figure 1).
Table 1 depicts the 17 studies, organized according to the author, informing year of publication, country of origin, design, target population and sample size.Prevalence, age range of the study population and criterion to define hearing loss are presented in Table 2.In all, the studies were conducted with a population of 367,330 (three hundred sixty-seven thousand three hundred and thirty) adolescents and young adults, whereas The procedures for hearing loss assessment in the studies of this review can be classified into three groups.In the first, there are four studies (23.5%) that used the WHO Ear and Hearing Disorders Survey Protocol for the Population-Based Survey of Prevalence and Causes of Deafness and Hearing Impairment and Other Ear Diseases, which proposes two stages of assessment, covering environmental noise measurement, application of questionnaire, otoscopy, pure-tone audiometry (over 4 years old) and impedanciometry.In the second group, there are nine surveys (52.9%) which use their own protocol for hearing loss assessment.Finally, the third group (23.6%), three studies were self-reported and evaluated with a tuning fork (Table 2).
Regarding the countries of origin for the research, six (35.3%) were in the USA, five (29.4%) in European countries, two (11.8%) in Brazil and four (23.6%) in other parts of the world (Table 1).Studies conducted in the United States of America had a mean prevalence of 14.7% (95% CI 13.5 to 16.0; I²: 66.7%; p = 0.01).The mean prevalence of those conducted in Europe was 5% (95% CI 1.6 to 13.5) and the ones developed in other countries was 10.2% (95% CI 5.0 to 17.0), and high heterogeneity was observed in these last two study groups, with I²: 99% and p value <0.001.
It should be noted that the two studies conducted in Brazil were quite distinctive household surveys.Béria and collaborators 18 carried out an audiological evaluation in children who were over four years old in a sample of the population from Canoas, RS, in 2003.They studied frequencies of 0.5 to 8 kHz and classified as mild hearing loss from 26 dB, detecting a prevalence of 7.1% in the age range from 10 to 19 years old.Cruz and collaborators 10 estimated hearing loss through interviews in which the hearing thresholds.Complementary observations of the inspection procedure, added to those disclosed by the tests, will surely reveal the audiological condition more accurately 34,35 .Shortage of papers in the established age group (15 to 24 years old) was a limitation of this study.Nevertheless, it was possible to identify studies which, even covering wide age ranges, showed prevalence by age strata.Another matter which is worthy to note is the variability in the prevalence of hearing loss in adolescents and young adults.It can be attributed, among other aspects, to different hearing loss identification models used in the studies, as well as the criteria for loss of degree classification, and also the diversity of the age group included in them.
In summary, in this meta-analysis, it was possible to observe a mean prevalence (meta-prevalence) of 12.0%, (95% CI 0.8 to 15) (Figure 2).It is worth noting the strong influence of both self-reported studies whose prevalence were 0.9% and 2.0%, while in the studies that measured hearing there was a variation from 7.1% to 23.8%.

CONCLUSION
Reading the papers, it was possible to identify a significant heterogeneity between the prevalence of self-reported hearing loss and the one measured by objective criteria.While in self-evaluation less than 2.0% of young people mentioned they had any deficit, studies conducting audiometry showed higher prevalence between 11.5 and 15.8%.
Currently, the use of earphones without any worry about duration or level of exposure is common.It is important to conduct studies to identify the temporary changes in the hearing threshold, in order to prevent permanent damage.
The necessity for research focusing on the population of adolescents and young adults is noteworthy, since they represent a part of the population who is susceptible to irreversible hearing loss due to high levels of indiscriminate non-occupational noise exposure intensity.
The four studies which evaluated hearing loss by the World Health Organization protocol presented a mean prevalence of 12.2% (95% CI 6.0 to 20.0) and considerable heterogeneity I²: 99.1%, p-value <0.001.Value similar to that observed in those eight ones which used their own protocols: 11.6% (95% CI 9.0 to 14.1; I²: 97.6% and p-value <0.001).
Five papers considered high frequency tonal threshold, classifying mild hearing loss from 16dB and observed prevalence near 15% [24][25][26][27][28][29] .Four studies considered the mean of three frequencies (i.e.500Hz, 1Khz and 2Khz) and established the mild hearing loss from 26dB 15,16,18,24 .The studies using the cutoff point for mild hearing loss from 26dB showed greater variability in the prevalence of hearing loss: 7.1 to 23.8% Studies with self-reported assessment of hearing loss showed lower prevalence than the ones that applied some type of measurement.This lower prevalence can be explained by the fact that high frequency hearing loss is barely perceptible, since it does not bring impact or obvious signs which impair daily activities 31 .This procedure underestimates hearing loss and can be a source of bias in interpreting results.
Differences found in the results may be explained, at least partially, by methodological aspects, especially those related to the definition criteria for hearing loss.The study of Abdel-Hamid and collaborators 15 used tuning fork test as a clinical procedure to diagnose hearing loss, while others used pure-tone audiometry, which is a test usually indicated for the determination of hearing thresholds (Table 2).Another potential source of variability in the prevalence of hearing loss is the subjectivity of pure-tone audiometry, which may underestimate or overestimate the true prevalence, giving the possibility of false positives 32 .
The conduction of otoscopy procedure as a careful inspection of the external auditory canal and tympanic membrane viewing is essential to control possible biases in audiometric results.Although it has no diagnostic aim, the test may provide information indicating infeasibility to conduct a pure-tone audiometry.Thus, absence of otoscopic data may suggest inaccuracy in the audiometric results due to the possible presence of foreign bodies and cerumen that would prevent a correct detection of pure-tone thresholds 33,34 .The papers that chose to conduct otoscopy, together with pure-tone audiometry and tympanometric results certainly bring a set of clinical information that can provide accurate data of

Table 1 )
. It is not easy to estimate the contribution of this variability, since there is not a clear relation, in the selected studies, between age groups and the prevalence of hearing loss.When people from several age ranges are compared, adolescents must be considered an age group who is susceptible to non-occupational noise-induced hearing loss and to the extra-auditory cultural, social and professional development.In this context, it is important to conduct studies to identify the temporary hearing threshold shifts, in order to prevent permanent damage.

Table 2 -Selected studies according to author, year of publication, age range, prevalence and criterion for hearing loss definition
Tung C-Y, Chao K-P.Effect of recreational noise exposure on hearing impairment among teenage students.Res Dev Disabil.2013;34(1):126-32. 5. Fligor BJ, Cox LC.Output levels of commercially available portable compact disc players and the potential risk to hearing.Ear Hear.2004;25(6):513-27.6. Biassoni EC, Serra MR, Richtert U, Joekes S, Yacci MR, Carignani JA et al.Recreational noise exposure and its effects on the hearing of adolescents.Part II: development of hearing disorders.Int J Audiol.