The occurrence of musculoskeletal complaints among professional musicians: a systematic review

Purpose This study gives a systematic overview of the literature on the occurrence of musculoskeletal complaints in professional instrumental musicians. Methods A systematic review. Nine literature databases were searched without time limits on June 25, 2015, also the complete index of the journal Medical Problems of Performing Artists (MPPA) until June 2015 (30;2) was searched, and citation tracking and reference checking of the selected articles were performed. The search consisted of the combination of three groups of keywords: musician (e.g., musician, violin, music student, instrument player) AND musculoskeletal (e.g., musculoskeletal, tendon, shoulder, arthritis) AND epidemiology (e.g., prevalence, incidence, occurrence). Results The initial literature search strategy resulted in 1258 potentially relevant articles. Finally, 21 articles describing 5424 musicians were included in this review. Point prevalences of musculoskeletal complaints in professional musicians range between 9 and 68 %; 12-month prevalences range between 41 and 93 %; and lifetime prevalences range between 62 and 93 %. Ten out of 12 studies show a higher prevalence of musculoskeletal complaints among women. Brass instrumentalists are reported to have the lowest prevalence rates of musculoskeletal complaints. The neck and shoulders are the anatomic areas most affected; the elbows are least affected. Although some information is reported concerning age, the high risk of bias in and between these studies makes it impossible to present reliable statements with respect to this. Conclusion Musculoskeletal symptoms are highly prevalent among musicians, especially among women instrumentalists. Future research concerning the epidemiology of musculoskeletal complaints among musicians should focus on associated risk factors and follow the current guidelines to optimize scientific quality.


Introduction
'There is no exercise, though never so healthful and innocent, but what may produce great disorders, if it is used with intemperance,' are the words of Bernardino Ramazzini, who was in 1713, the first to describe an overview of occupational diseases of musicians (Sataloff et al. 2010;Bejjani et al. 1996). Only at the end of the nineteenth century, a number of physicians turned their interest to some specific musicians' complaints like musicians' cramp. Tenotomies of the finger flexors were performed in order to improve finger independency among pianists (Sataloff et al. 2010). However, real interest in the health and wellbeing of musicians by medical practitioners, researchers and music professionals was developed since the 1980s.

3
This was reflected in a growing number of publications, journals, conferences, and organizations focused on the health of the performing artists (Sataloff et al. 2010;Bejjani et al. 1996). Nowadays, the level of knowledge on this topic and the necessary specialized healthcare is still in a developmental stage when compared to, for instance, sports medicine, and thus room for improvement remains.
Musculoskeletal complaints are one of the main medical problems among musicians (Guptill and Golem 2008;Hoppmann and Reid 1995;Heinan 2008). These complaints have considerable physical, psychological, social and financial impact on musicians (Spahn et al. 2001;Zaza et al. 1998). Impaired level of functioning at both work and daily activities at home due to these musculoskeletal complaints is reported in the majority, and sleep disturbances related to these complaints are reported in half of the professional musicians (Paarup et al. 2011;Kaneko et al. 2005). Most professional musicians will suffer from musculoskeletal complaints during their life; some of them will stop playing their instrument due to these complaints (Kaufman-Cohen and Ratzon 2011;Parry 2003;Kaneko et al. 2005).
Zaza published in 1998 a systematic review of incidence and prevalence of playing-related musculoskeletal complaints (Zaza 1998). In this study, 18 cross-sectional and cohort studies published between 1980 and 1996 were reviewed. Due to different definitions of musculoskeletal complaints, the point prevalence of the playing-related musculoskeletal disorders varied between 39 and 87 %. A development since this review is the introduction of the term playing-related musculoskeletal disorder (PRMD) , which aims to exclude minor irrelevant musculoskeletal symptoms experienced by musicians. Zaza defined PRMDs as personal, chronic and disabling health problems that affect the whole person, physically, emotionally, occupationally and socially ). However, the term PRMD is used in the literature of performing arts medicine without strictly following this definition. Recently, another review was published concerning pain prevalence in musicians (Silva et al. 2015). In this review, heterogenic studies are compared, with no distinction between professional and amateur musicians, impeding extrapolation of these results.
An up-to-date critical systematic review of the literature to assess prevalence rates of musculoskeletal complaints among musicians will indicate the extent of the problem, and a critical appraisal of the used prevalence rates and definitions of studied complaints will give an overview of the current science of musculoskeletal problems in musicians. Furthermore, subgroups with a higher prevalence can be identified. This may be helpful in the prevention of complaints due to the possibility to target prevention and interventions at these high risk groups. Therefore the objective of this systematic review is to give an overview of the prevalence of musculoskeletal complaints among professional instrumental musicians and to evaluate groups and localizations at risk.

Search strategy
A literature search was performed on June 25, 2015, using the following databases without time and language restrictions: PubMed, Embase, Web of Science, Cochrane, Cumulative Index to Nursing and Allied Health Literature (CINAHL), Academic Search Premier, PsycINFO, Scien-ceDirect and Lippincott Williams & Wilkins (LWW). The search consisted of the combination of three groups of keywords: musician (e.g., musician, violin, music student, instrument player) AND musculoskeletal (e.g., musculoskeletal, tendon, shoulder, arthritis) AND epidemiology (e.g., prevalence, incidence, occurrence). The complete search strategy is presented in "Appendix A". More over, the complete index of the journal Medical Problems of Performing Artists (MPPA) until June 2015 (30;2) was searched manually, and citation tracking and reference checking of the selected articles was performed.

Inclusion criteria
Articles were included if they fulfilled all of the following criteria: (1) the study had a cross-sectional, case-control or cohort design; (2) the study population consisted of adult (aged 18 or older) professional instrumental musicians and/or music academy students; the definition of professional was dependent on the definition of the original article. (3) The outcome measure reported was a clearly described prevalence rate of musculoskeletal complaints of the complete body or half of the body (at least upper extremities, back and neck) of musicians; (4) the article was published in a peer-reviewed scientific journal. If a subset of the total number of subjects included in a study met our inclusion criteria, the study was included only if the outcomes of the subset were assessed and reported independently.

Exclusion criteria
Studies with subjects aged 17 or younger were excluded. In case of unclear age limits, an indistinct description of the prevalence rate or questions concerning the professionalism of the study subjects, the authors were sent an e-mail. In case of non-response, the study was excluded. Case series that included less than 50 subjects were excluded.
Also studies reporting a prevalence of musculoskeletal complaints measured within a population visiting a healthcare professional were excluded. In case of a mixed study population, in which only a part of the study subjects met the inclusion criteria, authors were e-mailed and asked for split results. In case of a non-responding author or the inability of the author to present the relevant information, the study was excluded.

Study selection
Two reviewers (L.M.K., V.M.A.V.) independently performed the screening of title, abstract and full-text articles respectively, on eligibility. Disagreements in the selection process were resolved by consensus. When no consensus was reached, a third reviewer (B.M.A.H.) was consulted. In case of incomplete information in potentially relevant studies, the author was contacted by e-mail twice.

Data extraction
Two reviewers (L.M.K., V.M.A.V.) independently extracted the data from the included articles. General manuscript information (authors, title, year and journal) was collected. Information on the study population, sample size and response rate was listed. The prevalence rates of musculoskeletal complaints and specifications of these prevalence rates for differences in age, gender, occupation, localization and type of instrument were made. We also recorded whether the musculoskeletal complaints were playing related (yes/no). Disagreement between the reviewers was resolved by consensus.

Assessment of methodological quality
The methodological quality assessment was performed using a scoring system developed by Loney et al. (1998) and Shamliyan et al. (2010). This scoring system is specifically designed for studies reporting incidence and prevalence rates and consists of an eight-point checklist. Table 1 shows the quality criteria in eight categories: design and method; sampling; sample size; measurement criteria; bias; response and non-responders; outcomes; setting. A score ranging between zero (lowest score) and eight (highest score) indicates the quality of the included study (Table 1). Two independent reviewers (L.M.K., V.M.A.V.) assessed the quality of the studies. Disagreements were resolved by consensus. When no consensus was found, a third reviewer (B.M.A.H.) was consulted if the disagreement persisted.

Pooling of data
The aim was to pool the data if there would be sufficient homogeneity between the included studies.

Study selection
The initial literature search strategy resulted in 957 potentially relevant articles. Another 301 articles were identified after citation tracking and by checking the references of the selected articles. After the screening of title and abstract, Are validated questionnaires used?
5. Is the health outcome measured in an unbiased fashion? Is there a possible bias in the interpretation of the results?
6. Is the response rate adequate? Are the refusers described? >66.6 % response rate and dropouts described and compared with the study population 7. Are the estimates of prevalence or incidence given with confidence intervals and in detail by subgroup if appropriate?
8. Are the study subjects and the setting described in detail and similar to those of interest to you?
Are the sociodemographic characteristics adequately described?
Total 0-8 Points Scoring system: 0-4 points = low; 5-8 points = high 1 3 162 articles were considered eligible for inclusion and the full text was screened. Searching the MPPA database resulted in another 11 articles selected for full-text assessment. Finally, 21 articles, describing 17 studies and 5424 professional instrumental musicians, met our inclusion criteria and were included. Three study populations were reported on more than one article (Kok et al. 2013a, b;Fishbein et al. 1988;Fishbein 1988, 1989;Ackermann et al. 2012;Kenny and Ackermann 2015); results of these studies were pooled and presented as a single study. A flowchart of the inclusion and exclusion process is presented in Fig. 1. Articles without a clear description of age of the study population, (Fjellman-Wiklund and Chesky 2006;Kreutz et al. 2008;Hagglund and Jacobs 1996;Larsson et al. 1993;Miller et al. 2002;de Sousa et al. 2014) and articles lacking a clearly described type of prevalence rate (Mishra et al. 2013;Brandfonbrener 1997;Nemoto and Arino 2007;Mehrparvar et al. 2012;Hagglund and Jacobs 1996;Marques et al. 2003;Fjellman-Wiklund and Chesky 2006;Eller et al. 1992;Caldron et al. 1986;Wood 2014;Lopez and Farias 2013) were excluded. Also, studies with a mixed under aged population (Barton et al. 2008;Zetterberg et al. 1998;Caldron et al. 1986;Shields and Dockrell 2000;Wood 2014) or mixed occupation (Arnason et al. 2014;Wristen and Fountain 2013) (e.g., partly conductors or singers and not primary instrumental musicians) in which the authors were not able to present split data were excluded. Many articles derived from the UNT-MHS database (Chesky 2000) had to be excluded after e-mail contact with the main author due to the lack of a professional study population.  Table 3 gives an overview of the included studies. All included studies had a cross-sectional design. In fourteen articles symphony orchestra musicians were studied, in four articles music academy students, and in three other articles a mixed population of professional musicians and music academy students were studied. The included studies showed a variety of gender distributions, 26-79 % of the in the separate studies included participants were male. The studies were performed on a variety of continents: seven European, one Asian, five North American, two South American and two Oceanic studies were included. Response rates to questionnaires varied between 26 and 99 %; one study did not report a response rate (Arnason et al. 2014); and the number of participants of each separate study ranged between 59 and 2212.

Assessment of methodological quality
The results of the methodological quality assessment are presented in Table 2. Only two studies reported an adequate response rate (over 66.6 %) and an adequate description of the non-responders (Paarup et al. 2011;Engquist et al. 2004). Three studies used a validated outcome measure (Leaver et al. 2011;Kaufman-Cohen and Ratzon 2011;Engquist et al. 2004;Fotiadis et al. 2013), compared to 14 studies which used non-validated outcome measures or non-validated modifications of existing questionnaires.
Overall, the quality of the included studies was variable; however, many studies of low scientific quality were excluded from this review due to the strict exclusion criteria for reporting outcomes (Table 3).

Prevalence rates
A uniform definition of musculoskeletal complaints in the included studies was lacking; some authors used the definition of playing-related musculoskeletal complaints by Zaza et al. (1998) and Ackermann et al. (2012) ('any pain, weakness, numbness, tingling, or other symptoms that interfere with your ability to play your instrument at the level you are accustomed to'), whereas others (Engquist et al. 2004;Leaver et al. 2011;Paarup et al. 2011; Kaufman-Cohen and Ratzon 2011) used the questions based on the standardized Nordic Questionnaire (Leaver et al. 2011;Kuorinka et al. 1987;Fotiadis et al. 2013;Paarup et al. 2011;Engquist et al. 2004). In other studies, different descriptions such as '(joint-) pain,' or 'trouble' to describe the complaints were used (Kaneko et al. 2005;Kok et al. 2013a;Fishbein et al. 1988;Abreu-Ramos and Micheo 2007;Poolman et al. 2009).
Among the included studies, there was heterogeneity in the type of prevalence rates. Point prevalence, 12-month prevalence and life-time prevalence were most frequently reported. However, also 4-week prevalence, 'chronic' prevalence (with different definitions in the two reporting studies), and 3-month and 2-year prevalence were reported.
Two studies concerned all musculoskeletal complaints (without making a difference between playing-related or other complaints), whereas 12 studies measured only playing-related complaints. Three studies reported both playing-related and all musculoskeletal complaints. This variety of definitions of 'musculoskeletal complaints,' the heterogeneity of the reported prevalence types and the variability within study populations made it impossible to pool the data in this review.

Prevalence rates
Reported point prevalence rates of musculoskeletal complaints, presented in Table 4, varied from 57 to 68 % for all musculoskeletal complaints, and from 9 to 68 % for playing-related complaints. Non-playing-related 12-month prevalence ranged between 86 and 89 %, and playingrelated 12-month prevalence ranged between 41 and 93 %. Playing-related lifetime prevalence ranged between 62 and 93 %. No study reported non-playing-related lifetime prevalence.

Gender
Ten out of 12 studies comparing the gender of the professional musicians showed a higher prevalence of musculoskeletal complaints among women. One study (Kaufman-Cohen and Ratzon 2011) only stated 'no significant difference' without presenting the data, and another study (Davies and Mangion 2002) reported a higher prevalence among female compared with male strings players, but a lower prevalence among females playing another instrument. However, no exact data were given in this study. Table 5 shows the results of the gender-specific prevalence rates.

Occupation
There were no studies that compared prevalence rates of musculoskeletal complaints between different occupational groups (e.g., orchestral musicians, music teachers, music academy students). Kok et al. (2013a, b) reported a point prevalence of 63 % of musculoskeletal complaints among music academy students. The latter was in concordance with Kaneko et al. (2005) and Engquist et al. (2004) who reported prevalence rates of 68 and 61 % respectively, in orchestra musicians.      Also the 12-month prevalence of 89 % among music academy students in the study of Kok et al. was comparable to the prevalence rates of orchestra musicians of Leaver et al. (2011) and Paarup et al. (2011), 86 and 88 % respectively. No information was presented in the included articles concerning prevalence rates between different occupations among professional musicians; e.g., teachers, chamber musicians, soloists and orchestra musicians.

Instrument
In addition to the above-mentioned heterogeneity in the definition of measured complaints and the type of prevalence reported in each study, heterogeneity in the grouping of instrumentalists and the presentation of differences between these instrument groups were presented in the included studies. Some authors reported no total prevalence rate split for instrument groups, only body-area-specific prevalence rate split for instrument groups (Leaver et al. 2011;Paarup et al. 2011;Roach et al. 1994). As there is a possibility to have multiple complaints, these numbers could not be summed up. An overview of the reported prevalence rates for each instrument group is presented in Table 6. Overall, no specific instrument group had an evidently higher prevalence rate of musculoskeletal complaints. However, brass instrumentalists were reported to have the lowest prevalence rates of musculoskeletal complaints (Arnason et al. 2014;Leaver et al. 2011;Paarup et al. 2011;Abreu-Ramos and Micheo 2007;Kaneko et al. 2005;Roach et al. 1994;Kok et al. 2013a;Fishbein et al. 1988;Ackermann et al. 2012;Steinmetz et al. 2015).

Age
One study compares lifetime prevalence rates of musculoskeletal complaints among age groups (Abreu-Ramos and Micheo 2007). The highest prevalence rates were reported in the highest (50-61 years; 91 %) and youngest (22-29, 83 %) age groups.

Anatomic region
Above-mentioned differences in the reporting of complaints are also reflected in the heterogeneity of studied body areas. The number of reported body areas differed from four (e.g., neck, shoulder (both/left/right), fingers (each separate, or in general) up to 32. This high variability between affected anatomical areas (i.e., heterogeneity in location of complaints) made comparison between the included studies difficult, since multiple complaints at several anatomic regions can be present, as well as radiation of these complaints to different anatomical regions. In Table 7, the prevalence rates for each anatomic region are     presented. Overall, the neck and shoulders were most frequently affected, and the elbows had the lowest prevalence rate of musculoskeletal complaints. No differences between left and right side of the body were evident.

Discussion
This systematic review focused on the prevalence of musculoskeletal complaints among professional musicians. In the included articles, there was a wide variability in the definition of these complaints as well as on the outcome measures used. The point prevalence of all musculoskeletal complaints ranged between 9 and 68 % and for playingrelated musculoskeletal complaints between 9 and 68 %. Twelve-month prevalence ranged between 86 and 89 %, and playing-related 12-month prevalence ranged between 41 and 93 %. Playing-related lifetime prevalence ranged between 62 and 93 %. In most studies, women have a higher prevalence of complaints compared to men.

Limitations of this study
Due to heterogeneity on several aspects between the studies in this systematic review, pooling of study data was not possible. Since the critical review of Zaza et al. (1998), more than a 100 new articles describing musculoskeletal complaints among musicians were published. Of these articles, 12 were included in this review. Many of these recently published articles lack essential methodological information (e.g. biased or non-described selection of participants, lack of reporting a response rate or a clear cut definition of the measured complaints). Also, the results section is often lacking important information (e.g. location as well as duration of the complaints). Furthermore, selection bias is often present in these studies. The latter is exemplified by missing general baseline information, like age and gender on the study subjects as well as which patients are selected to be included in the study and what the loss of follow-up is (i.e., response rate) (Eller et al. 1992;Fjellman-Wiklund and Chesky 2006;Kreutz et al. 2008;Hagglund and Jacobs 1996;Larsson et al. 1993;Miller et al. 2002;Brandfonbrener 1997;Nemoto and Arino 2007;Mehrparvar et al. 2012;Marques et al. 2003;Caldron et al. 1986;Arnason et al. 2014;de Sousa et al. 2014).
As described in the methods section, all articles lacking a clear description of the study population or a measured prevalence rate were excluded from this review. Therefore, the quality of the included studies in this review is generally high compared with the overall performing arts medicine literature. This is confirmed by the used methodological quality score on which 17 out of 21 studies score high. Another limitation of this study is the lack of 'non-classical' professional musicians, e.g., musicians playing in a marching or pop/rock band. As these musicians have both another musculoskeletal load (e.g. standing performance instead of sitting) and another lifestyle, they possibly have other musculoskeletal problems compared to the classically trained musicians.

Musculoskeletal complaints in musicians and subgroups at risk
We found that females have a higher prevalence of musculoskeletal complaints when compared with men, and this is in line with the literature of musculoskeletal complaints in the general population: female gender is a known risk factor for development of these complaints (Picavet and Schouten 2003).
Although comparison of the studies describing prevalence rates in music academy students and professional musicians was difficult due to heterogeneity, no evident difference in prevalence rate between music academy students and professional orchestra musicians was found.
Comparison of prevalence rates of musculoskeletal complaints between musicians who play different instruments did not show a specific instrumental group with an evidently highest prevalence rate, although brass instrument players had the lowest prevalence. It should be noted that some of the musicians play multiple instruments, where all included studies describe only the main instrument. Also, the instrument categories used consisted of instruments which are varying in size and playing position and technique. For example, the category strings consists of violin, viola (somewhat larger and heavier compared to the violin), cello and base players and in some studies even guitar players. The playing position of a base player is completely different compared with a violinist, and since the instrument is larger, a sitting position is used, and the repertoire of the base player (heavy, slow and often repetitive) is different compared with the fast and virtuoso repertoire of the violin. Thus, since the included articles combine the prevalence rates in groups of players, no distinction between subgroups of string players can be made.
No valid conclusion can be drawn from this review concerning the relation between age and musculoskeletal complaints among professional musicians. Only one study compared age groups, but this study used a lifetime prevalence rate, and the risk of recall bias is high when using lifetime prevalence rates (Moffitt et al. 2010). However, musculoskeletal complaints in the general population are most frequent among subjects in the fifth, sixth and seventh decade of their life (Picavet and Schouten 2003;Huisstede et al. 2008Huisstede et al. , 2006. As musicians pass through the same aging process, it is supposed that the highest prevalence of musculoskeletal complaints among them would be the same compared to the general population. However, there might be a 'healthy player effect,' in which musicians with severe musculoskeletal complaints quit their career before reaching this age. Therefore, musicians could have another distribution of musculoskeletal complaints in age groups compared to the general population.

PRMDs/non-PRMDs
The term 'PRMDs' was introduced to evaluate musculoskeletal symptoms which interfere with the ability to play the instrument ). Since then, many studies evaluated these playing-related symptoms instead of evaluating all musculoskeletal symptoms, thereby excluding minor symptoms (Zaza and Farewell 1997;Davies and Mangion 2002;O'Neill et al. 2001;Ackermann et al. 2012). The use of this term has an important advantage; symptoms without impact on the musician (and therefore irrelevant symptoms) are excluded. However, the comparison of musicians with non-musicians is difficult with this definition. Besides, although Zaza et al. (1998)

Recommendations for future research
We recommend that future research should aim for a higher level of methodological quality to contribute to the existing knowledge of the occurrence and risk factors for musicians' musculoskeletal complaints. A minimum requirement is data on the included cohort, a brief definition of the measured musculoskeletal complaints (i.e. anatomic area, radiation etc.), data on loss of follow-up and the use of validated outcome measures. Focus should be on selecting subjects while avoiding bias (adequate response rate, describing non-responders and selection procedure), using adequate and validated instruments for measuring all outcomes; Using the DASH, SF-36, Michigan hand score, Nordic Questionnaire etc., has strong preference above using a non-validated self-made or adapted (modified existing, and not re-validated) questionnaire (Poolman et al. 2009). Scientific guidelines, for example the STROBE or IDEAL and NOS are recommended for increasing the quality of future studies (Elm et al. 2007;McCulloch et al. 2009).

Conclusion
Musculoskeletal symptoms are highly prevalent among musicians, especially among women. In contrast to the literature on musculoskeletal complaints in the general population, evidence is scarce concerning prevalence rates in subgroups of age or occupation. Future research concerning the epidemiology of musculoskeletal complaints among musicians should focus on associated risk factors and follow the current guidelines (McCulloch et al. 2009;Elm et al. 2007) to optimize scientific quality.

Conflict of interest None.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.