Noise-induced hearing loss is a widespread condition in occupational health (1). The WHO attributes 16% of global hearing loss to occupational noise exposure (2). In Germany, noise-induced hearing loss is the most commonly recognized occupational disease (3) and is often a result of chronic occupational noise exposure and insufficient protection measures (4). In addition, musicians can be affected, even though music is a very special type of noise (5). Occupational examinations reveal an elevated risk of hearing loss for musicians playing in an orchestra: 8 h exposure levels may exceed L_Aeq8h 85 dB(A) or peak values of 137 dB(C), depending on the instrument and orchestra position (6). The law mandates that at these noise levels, employers must provide protective measures (7, 8), which are often inappropriate for musical activities. For instance, hearing protection devices can distort the sound frequency range (9) and lead to the occlusion effect, which affects musicians’ perception of the music (10). Thus, the use of hearing protection devices is low among musicians (11). Partition and sound baffles may worsen sound exposure through reflection effects or by causing inhomogeneous sound dispersion (12). Substituting loud sections of music is not an option, and administrative measures to reduce the volume during rehearsal or concerts are limited.

Given these facts, it is important to determine whether musicians have an increased prevalence of music-induced hearing loss. Previous studies have provided an overview of the topic but lack a systematic literature search. For instance, Marquard and Schäcke reported in detail in 1998 on the heterogeneity of the published studies assessing musicians’ hearing loss (13). They found major differences in outcome parameters, methodical approaches, examination conditions, and participant groups. More recent reviews assessed a broad range of musicians’ diseases and briefly reported on hearing loss (14, 15). Other reviews focused on noise as the factor inducing hearing loss in musicians (16, 17). Pharmaceutical interventions for the treatment of musicians’ hearing loss have been proposed by Wartinger et al. (18). Behar et al. reported on studies using pure tone audiometry but focused on the technique of noise measurement (19). They provided recommendations about noise measurement but did not draw conclusions about the prevalence of musicians’ hearing loss. All of the mentioned studies are not systematic reviews. The only systematic review on the topic is by Di Stadio et al., published in 2018 (20). This thorough review involved a subgroup analysis of pop/rock and classical music musicians. The review included 41 studies and pooled the results from a sample of 4,618 professional musicians. The authors reported a prevalence of musicians’ hearing loss of 63.5% for pop/rock musicians and 32.8% for classical music performers in the frequency range of 3,000–6,000 Hz. A limitation of the review was the heterogeneity of the retrieved studies. The pooling of the results includes a high risk of bias since studies with low quality have a high impact. Furthermore, the proportion of affected individuals relative to the original sample size (prevalence) can be obscured by pooling the results, as well as by grouping the results into different subgroups containing different instrumentalists. With a new methodical approach that preserves the original prevalence findings, we sought to update the research question.

Furthermore, it is important to consider the criteria that distinguish a well-conducted study and the publication of meaningful data. This starts with the design of the study (e.g., cohort size) and extends to the execution of audiometry (e.g., environment, noise breaks, and use of a standardized audiometer), evaluation of the raw data (e.g., age correction or at least its reporting), and preparation of the data for publication (e.g., reporting hearing loss in dB and by frequency).

We did not aim to compare occupational risk assessments; rather, our goal was to determine the prevalence of music-induced hearing loss and its characteristics among musicians.

We conducted a literature search following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines (21), using the search terms (hearing loss OR hearing impairment OR hearing difficulties OR acoustic trauma) AND (musician). This search was performed between 14 August 2023 and 2 January 2025, for a final update. We searched the databases such as MEDLINE, Embase, Cochrane Library, and Google Scholar. For the latter, we restricted the search results to a total of 300 entries. The abstracts and titles from the Google Scholar result list were excluded from HTML files using Wolfram Mathematica 13.1. URLs were used when digital object identifiers (DOIs) could not be retrieved. Each Google Scholar entry was added to the screening list. We also included theses and comparable university works to reduce the risk of bias. We applied no filters and included all years of publication. The review was not registered, and the protocol was not published in advance.

The inclusion criteria were as follows: an examination of musicians (mostly professional), and majority age of participants (population), exposure to music performance (exposure), and original research that included the use of pure tone audiometry (outcome).

The exclusion criteria included singers as the study population, music exposure in a leisure context, and publication types such as case report studies, systematic reviews, reports, grey literature, and languages other than English or German. Computational translations for other languages yielded inappropriate results. The study design was not a selection criterion as long as audiometry data were reported for musicians. A control group was not required since we focused on the prevalence of noise-induced hearing loss.

During the review process, we never used automation tools or citation tracking tools. The screening was performed independently by the two authors as reviewers, using xlsx-files. First, the abstracts were screened. The unclear and deviating titles were discussed. In the second step, the full texts were screened in the same manner. The reference lists were not searched systematically, and we did not add any manual entries. The flowchart is presented in Figure 1.

Data extraction was conducted simultaneously and under the constant observation of the other reviewer. The data extraction was based on the following categories: musical genre, study population, number of participants (including the female-to-male ratio) age range of study population, minimum years of musical practice, inclusion and exclusion criteria of the study, instruments used (pure tone audiometry), and results. The results were further specified by the findings (normal hearing, notch, and hearing loss), the affected side of the ear, affected participants or subgroups, and the proportion of the affected participants to the study population. Finally, we assessed the data reporting to determine whether it was detailed and sufficient or poorly done with regard to the specified criteria, including reporting of frequency range, hearing loss in dB, and age correction. In addition, the study design (cross-sectional vs. longitudinal), the kind of data acquisition, and the reporting of raw data were also evaluated. Assessing the risk of bias was not necessary because of the descriptive epidemiological approach. We did not aim to conduct statistical subgroup analyses a priori but focused on the results of the musicians’ pure tone audiometry, as described previously using the PEO criteria. As there are no formal quality assessment tools for pure tone audiometry, we decided to limit the results to a descriptive evaluation based on the recommendations of the American Speech–Language–Hearing Association (ASHA). Except for the criterion “age correction,” which was added post hoc, all extracted categories were defined primarily.

Studies that included only healthy participants for another primary research issue (e.g., tinnitus) were not included in the final analysis since the prevalence of hearing loss may be biased. Furthermore, the findings that were solely based on comparisons (subgroup, ear side) were excluded since they did not contain prevalence data for the study population.

To conduct a systematic analysis, we grouped the findings into four musical genres: classical music, rock/pop/jazz, traditional music, and military and marching band music. The unreported studies were assigned to the group “unknown.”

For statistics and graphs, we used Wolfram Mathematica 13.1. Descriptive statistics consisted of stacked histograms showing the distribution of quantitative data (number of participants and years of practice). For the distribution of sex, we calculated the percentage values and plotted them in a sorted bar chart. The pure tone audiometry results were categorized into four groups as follows: (1) values with frequency and hearing loss reporting, (2) values with frequency but without hearing loss in dB reporting, (3) values without frequency but with dB hearing loss reporting, and (4) values with neither frequency nor hearing loss reporting. All plot data were categorized by genre, and the resulting counts were collected. A color scale was used for visual coding of the count numbers. These values were plotted in a “conventional” audiogram layout. This approach ensured that no data had to be transformed and that partially missing data could also be reported.

For the visualization of the proportion of the affected persons, we calculated the percentage values and plotted sorted bar charts with opacity decreasing in relation to the maximum study population. Inferential statistics were employed where necessary, using Wolfram Mathematica 13.1. To compare the distribution of categorical variables with more than 50 values (e.g., sex distribution), we used Pearson’s chi-squared test, with α = 0.05.

The peer review process with two blinded reviewers yielded 79 studies, according to the PRISMA guidelines. The studies were heterogeneous in terms of the genre, publication of the results, and audiogram data corrected for age. The majority of the retrieved studies focused on young, male musicians with little exposure time. Older or retired musicians, as well as women, were underrepresented or missing. The summary table can be found as an online supplementary material in our repository (33).

For the classical genre, some studies found no hearing loss in their study population, while others found notch configurations or hearing loss >15 dB in the frequency range of 4,000–6,000 Hz. This indicates a higher risk of hearing loss for classical musicians. The prevalence of notch configurations and hearing loss varied between 5 and 70%. A total of 20–60% of rock, pop, and jazz musicians tended to have a higher risk of hearing loss >20 dB in the range of 3,000–8,000 Hz, as expected. The data for military music and traditional music were limited, although an increased risk of hearing loss was also found in the 4,000–6,000 Hz range.

These findings are consistent with an analysis of diagnoses from 2,227 musicians based on data from 7 million German health-insured individuals, which found a hazard ratio of 3.51 (95% Confidence Interval: 2.82 to 4.21) for noise-induced hearing loss in musicians compared to non-insured musicians (5). Concerning the prevalence of noise-induced hearing loss in musicians, a report published in 2009 showed that 58% of musicians in the classical genre and up to 49% in the rock/pop genre were affected (112). The systematic review by Di Stadio et al., presented in the Introduction section, reported hearing loss in 63.5% of rock/pop musicians and 32.8% of classical music performers. Instead of focusing on a mean value, we adopted a holistic approach for reporting the range of the prevalence findings. The findings from the other studies are consistent with the range of our results, as shown in Figure 9 through Figure 12.

Notch configurations and hearing loss in the high-frequency range of 4,000–6,000 Hz are characteristic of noise-induced hearing loss (113–115). Speech-frequency hearing impairment in the range of 500–4,000 Hz is much more common in the general population, with a prevalence of 14.1% in the USA (116). This study also assessed the prevalence of bilateral hearing impairment in relation to occupational noise exposure, defined as exposure “at work to loud sounds or noise for four or more hours, several days a week”. The findings were as follows: no exposure: 14% (95% CI: 13–16), exposure to noise “so loud that they had to raise their voice to be heard” for more than 5 years: 28% (95% CI: 21–37), and exposure to noise “so loud that they had to shout to be heard” for more than 5 years: 43% (95% CI: 35–51). Compared to the retrieved studies, which mostly focused on exposure for more than 5 years, musicians may have an increased prevalence of hearing loss, especially in the rock, pop, and jazz genres (20–60% vs. 14.1%). This prevalence is within the range observed in occupationally exposed individuals: 28 and 43%. However, in our results, the prevalence of hearing loss among musicians in the classical music genre was too broad to draw meaningful conclusions (5–70%). All the studies identified in this review were original works investigating musicians. Only a few had sample sizes >300 and a cohort study design, such as the study by Karlson et al. (107). As shown in Figure 9 through Figure 12, the reporting of hearing loss prevalence depends on different factors. These include the composition of the affected group, which may consist of different instrumentalists; the reporting of audiometry results, which may include dB hearing loss and frequency range; and the type of age correction applied. To discuss these factors, we need to take a closer look at some publications from the classical genre that provide detailed information about them.

Obeling and Poulsen found no signs of hearing loss in 57 musicians from four Danish orchestras (37). They corrected the audiograms using ISO 1999, taking into account the number of years of musical practice, playing hours per week, and the average sound level. The authors published averaged audiogram plots with standard deviations. They reported results for ear-side and instrument groups but did not include a comparison by sex. The authors emphasized that the data were not representative since the sample size was small. Assuming 90 musicians in a symphony orchestra, the participation rate in this study was 15%.

Toppila et al. assessed 63 musicians from 4 symphony orchestras, resulting in a participation rate of approximately 17% (41). The authors corrected for age, noise exposure, and sex using ISO 1999-1990. The audiometry results for both ears were plotted with 95% confidence intervals. Box and whisker charts reported the corresponding z-scores for comparison between a noise-exposed and a non-exposed standard population. They did not find any differences in the musician group compared to the non-exposed population. The authors compared the mean differences in hearing loss between two subgroups: musicians with high exposure (L_ex > 100) and musicians with low exposure (L_ex < 100). The hearing loss values were significantly higher in the range of 1,000–6,000 Hz for the high-exposure subgroup.

Wegner et al. assessed 40 classical music instrumentalists and reported a participation rate of 30% (48). They reported averaged hearing levels with standard deviations for both ears for the instrumental groups, as well as the use of hearing protection devices. The values were age-corrected based on a preceding study from 1967. Sex comparison was not reported. The authors found normal hearing in the sample.

Eaton and Gillis found notch configurations in the high-frequency range in 13 of 53 musicians from a Canadian symphony orchestra (46). The participation rate was approximately 50%. The audiogram data were corrected using ISO 1999-1990 (as described previously). The averaged audiometry results were compared for sex and instrument groups. Differences in the ear side could not be found.

Kähäri et al. had a sample size of 140 participants from two Swedish orchestras, with a participation rate of 79% (38). They reported mean values with standard deviations, as well as median and range for both ears. The plots show the 10th and 90th percentiles, as well as groupings by musical instrument, sex, and age. The data were not corrected for age, but the age groups are reported in detail. The findings indicated no severe hearing loss among musicians but highlighted a more frequent notch configuration in the men’s audiogram and slightly worse hearing in the percussion and woodwind players.

Participation rates are often low, which may introduce a bias, with individuals experiencing hearing loss less likely to participate in the study. Reporting averaged hearing loss values carries the risk of underestimating subgroup differences and may result in the loss of information about the distribution of the original audiogram data. Age correction often relies on ISO 1999-1990 references (37, 41, 46), and noise exposure must be calculated beforehand (37). The heterogeneity of assessments demands standardized methods and prospective study designs. Longitudinal studies are completely lacking, making it impossible to assess the lifetime prevalence of hearing loss among musicians. Prospective multicenter studies are needed to address this research question.

Knowledge about music-induced hearing loss is crucial since hearing ability is essential for a musician’s work. Occupational hearing loss often results from continuous exposure and may go unnoticed by the affected person (12, 117). This highlights the importance of regular audiometric examinations for musicians, in accordance with the laws of the respective country. Only through this approach can music-induced hearing loss be detected and preventive measures be implemented. For the latter, systematic and scientific evaluation is needed to avoid accidental harm. Organizational measures should also be considered since they are easy to implement. For example, reducing the playing volume during rehearsals could help mitigate risk. To avoid temporary threshold shifts from prolonged exposure, loud passages could be played at the end of a rehearsal. After rehearsal, sufficient recuperation time should be ensured. Music presents a complex challenge that rarely allows for the use of conventional preventive hearing protection measures typical in occupational health.