Introduction

To a great extent, professional musicians rely on their physical and mental health to guarantee their vocational and artistic positions, meet high audience demands or succeed in competitive settings. Even minor complaints may impair the precision of motion sequences and musical technique, thus creating a major threat to the artist’s existence.

The discipline of musicians’ medicine dates back to the 18th century (Ramazzini 1705). In the 1920s, Kurt Singer first systematically described symptoms of musicians’ vocational diseases and their treatment (Singer and Lakond 1932; Harman 1993). Currently, musicians’ medicine is dedicated to the prevention, diagnosis and therapy of health problems which may arise or have arisen as a result of making music or which have an effect on making music (Spahn et al. 2011).

In clinical practice, musculoskeletal and mental problems, especially performance anxiety, are very common amongst professional musicians (Fishbein et al. 1988). Up to now, several narrative and systematic reviews exist that provide data on a wide range of playing-related symptoms and diseases, with broadly varying prevalence rates (Harman 1982; Zaza 1998; Milan 1996; Zuskin et al. 2005). The varying prevalence rates may mainly be caused by the lack of a precise definition of playing-related symptoms and diseases as well as a lack of coordinated research.

Reviews of publications including patients with musculoskeletal complaints and disorders (MCD) used different definitions of musculoskeletal disorders or simply referenced “overuse syndrome” (Hoppmann and Patrone 1989; Bejjani et al. 1996). In 1998, Zaza (1998) introduced the term “playing-related musculoskeletal disorders” (PRMD), which aggregated the various musculoskeletal disorders while assuming a common etiological factor. The prevalence of PRMD in musicians was thus deemed to be comparable to vocation-related musculoskeletal disorders in other professions.

The most recent systematic review by Bragge et al. (2006) furthered the use of PRMD as an aggregate term for overuse syndrome, repetitive strain injuries or cumulative trauma disorders. The review reported a prevalence ranging between 26 and 93%. Further systematic reviews added only limited information, as they did not use predefined review protocols or used a narrow search strategy (Wu 2007; Moraes and Antunes 2012). Furthermore, previous reviews did not focus on the methodological quality of the included studies.

Since various musculoskeletal disorders were inconsistently summarized under the term PRMD and the methodological quality of included publications was not assessed in previous reviews, we aimed to perform a comprehensive systematic review including published literature without language restrictions, based on an elaborated study protocol, to assess the prevalence, risk factors, prevention and effectiveness of MCD treatment in professional musicians, including the assessment of the methodological quality of the included studies.

Methods

The research methods and reporting of this study followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines (Liberati et al. 2009; Moher et al. 2009) and the recommendations of the Cochrane Collaboration (Green and Higgins 2011). Prior to conducting the review, a study protocol was prepared by all of the review authors, placing special emphasis on study selection, data extraction and quality assessment.

Criteria for considering studies for the current review

Types of studies

Observational studies (case–control studies, cohort studies, cross-sectional studies), intervention studies (controlled clinical trials and pre–post intervention studies without a control group), case reports and case series reporting clinical interventions published in peer-reviewed journals without language or publication date restrictions were included in this review. Studies published in non-peer-reviewed journals, theses, and gray literature were excluded.

Types of participants

Participants in the included studies were male or female professional musicians of all musical genres, including music teachers, instrument teachers and music students in higher education institutions (for example, universities, colleges, conservatories). Studies examining mixed professional populations (amateur, semi-professional/professional) or mixed artistic populations (musicians, actors, dancers) were included only if results were reported for the respective subgroups. When the professional status of the participants was not clearly defined in the publication, a consensus of the reviewing authors was obtained. For studies involving high school music students, participants had to be of at least 16 years of age (defined in this review as “adult”) to be included. Publications on mixed populations (children, adolescents, adults) were included only if results were reported for the respective subgroups. The included studies investigated MCD that was potentially caused by or thought to be related to practicing or performing music. This also included dental and jaw diseases, myofascial pain syndromes, craniomandibular dysfunctions, shoulder belt compression syndromes (like thoracic outlet syndrome and congestion syndrome of the upper thoracic aperture) and percussion hemoglobinuria. Studies lacking relevant information on the different types of participants were excluded.

Types of interventions

Studies investigating the effectiveness or efficacy of any type of clinical intervention, including interventions using complementary and integrated medicine in the defined study population, were considered.

Types of comparisons

A control group comparison was not required to meet the eligibility criteria.

Types of outcome measures

Studies were considered if the primary outcomes included the prevalence, incidence or other information about the prevalence of MCD or risk factors for MCD or clinical treatment effects due to an intervention (preventive, therapeutic or rehabilitative). Studies providing only non-clinical treatment effects or validating methods of measurements were not included.

Data sources and searches

The electronic databases MEDLINE and EMBASE via OvidSP and via EbscoHost CINAHL, PsycArticles, PsycInfo and ERIC were searched between January 6th, 2015 and December 7th, 2017, with no limit on the publication date.

Hand searches were performed as follows: the scientific journal Medical Problems of Performing Artists (MPPA, until volume 24, issue 4 December 2009, after in MEDLINE) between February 13th and 27th, 2015 and actualized on December 7th, 2017 as well as the German journal Fachzeitschrift Musikphysiologie und Musikermedizin (FMM, years 1994–1999 were excluded, because online access was not available), the official scientific publication of the German Association for Music Physiology and Musicians’ Medicine (DGfMM), between February 28th and March 8th 2015 and actualized on December 7th, 2017. Reference lists of seven identified systematic reviews (Zaza 1998; Bragge et al. 2006; Wu 2007; Baadjou et al. 2016; Jacukowicz 2016; Kok et al. 2016; Vervainioti and Alexopoulos 2015) were searched for further studies. The search strategy followed three guidelines: the study population, investigated MCD and study design. The search strategy is shown in Appendix 1.

Data collection and analyses

Study selection

The titles and abstracts of the identified studies were screened for eligibility by two authors (KN, GR). Studies not meeting the inclusion criteria were excluded. The remaining studies were evaluated for inclusion via full-text reviews by two of the four authors (KN, GR, AS, AB). During the full-text review, a predefined checklist form was completed as reported in “Assessment of study quality and dealing with missing data”. For articles not published in English or German, a translated summary was examined for eligibility criteria. Discrepancies in the study selection between authors were resolved in a consensus conference.

Data extraction

Data extraction was performed by two of the four authors (KN, GR, AS, AB) using standardized data extraction forms as reported below. Discrepancies in the data extraction between the reviewing authors were resolved by discussion and consensus.

Assessment of study quality and dealing with missing data

Depending on the study design, the quality of studies can be assessed using various checklists or scores. A standard assessment tool is the Study Quality Assessment Tool of the National Heart, Lung, and Blood Institute (NHLBI). It provides a thorough assessment of the quality of studies in all medical disciplines. It can be applied to observational studies such as intervention studies and seemed appropriate to be extended by a scoring system for comparison purposes in this review.

To apply comparable quality assessment tools to all of the included study designs, the reviewing authors developed a modified version of the “Study Quality Assessment Tool” from the National Heart, Lung, and Blood Institute. The “Quality Assessment of Observational Cohort and Cross-Sectional Studies,” “Quality Assessment of Before-After (Pre–Post) Studies With No Control Group” and the “Quality Assessment of Controlled Intervention Studies” (NHLBI) were adapted by adding items to the quality assessment tools about the Critical Appraisal Skills Program (CASP), the “CASP-Checklists” (CASP 2013) and the “Methodology Checklists” of the Scottish Intercollegiate Guidelines Network (SIGN). The instructions for application were predefined and adapted in a consensus by all authors. A scoring system was implemented in which one point was given for each item on the form that was fulfilled by the study and one point was subtracted for each item that was not fulfilled. If an item was not applicable to a study, zero points were given. If a study did not report on the requested item, one point was subtracted for that item. The total score was calculated by adding the individual points without weighting for single items. The total number of possible points varied for each evaluation instrument depending on the type of study. Case–control studies could reach a maximum of 14 points, cohort studies and cross-sectional studies could reach a maximum of 16 points, controlled intervention studies could reach a maximum of 18 points and pre–post studies without a control group could reach a maximum of 15 points (Appendix 2). Total scores were not comparable across different types of study designs. The described modified quality assessment tools were used for the quality assessments of the individual studies by two of the four authors (KN, GR, AS, AB). Disagreements over study quality assessment and scoring were resolved by discussion and consensus. If multiple publications were produced from one study, only one quality assessment was performed based on the publication we considered to be the most comprehensive and with the highest methodological quality.

Data synthesis

The extracted information included the (1) authors, (2) date of publication, (3) population(s) studied, (4) number of participants, (5) exposure or interventions relevant to the review questions, (6) randomization status; (7) outcomes and (8) results. If these data were not provided, they were marked as missing. The results of individual studies were reported as frequencies in percentages, effect sizes, mean values with standard deviations or standard errors, significance values, odds ratios and confidence intervals. With the exception of percentages, no calculations were made based on the values provided. If a study did not provide quantitative data in numbers, the narratively described results were used.

Results

Overview of selected studies

The search strategy identified 2074 articles. Figure 1 depicts the results of the individual steps of the selection process. In total, 109 articles were included in the study, 28 of which were case studies and 81 of which were articles for quality assessment. The 81 articles assessed for methodological quality included 1 case–control study, 6 cohort studies, 62 cross-sectional studies, and 12 interventional studies (9 controlled intervention studies and 3 pre–post studies without control groups).

Fig. 1
figure 1

Study selection process

Full size image

The characteristics of the included studies are shown in Tables 1 through 6. Across all study types, we found that the different terms used and the definitions of outcome parameters significantly limited the comparability of the included studies. In the “outcomes” columns in Tables 1 through 5, the individual terms used by the studies are cited. Even the abbreviation “PRMD” is used and defined heterogeneously.

Table 1 Characteristics and results of the included studies, case–control study
Full size table

The term “PRMD” as “performance-related musculoskeletal disorders” was used in seven publications (Ackermann et al. 2002b, 2011, 2012; Chan et al. 2013; Chan et al. 2014; Khalsa and Cope 2006; Khalsa et al. 2009). Among these, two intervention studies defined performance-related musculoskeletal disorders according to Zaza and Farewell as “any pain, weakness, numbness, tingling or any other symptoms that interfere with your ability to play your instrument at the level you are accustomed to. This definition does not include transient aches or pains” (Ackermann et al. 2002b; Chan et al. 2014; Zaza and Farewell 1997). One study defined the same term as “a musculoskeletal disorder was considered performance-related if the injury occurred during or immediately after playing and the musician specified that playing the instrument was the main contributor to their injury“(Chan et al. 2013). The term “PRMD” as “performance-related musculoskeletal pain disorders” was applied in three other publications (Kenny and Ackermann 2015; Ackermann et al. 2012; Kenny et al. 2016) that used the definition based on Zaza and Farewell (Zaza and Farewell 1997), as mentioned above. The term “PRMD” as “playing-related musculoskeletal disorders” was used in ten studies (Steinmetz et al. 2012; Arnason et al. 2014; Kim et al. 2012; Kaufman-Cohen and Ratzon 2011; Mishra et al. 2013; Kochem and Silva 2017; Monaco et al. 2012; Sousa et al. 2016; de Greef et al. 2003; Rickert et al. 2012). The definitions within the studies were heterogeneous; for example, in one study (Rickert et al. 2012), the term “PRMD” was used without further definition but was used synonymously with “injury” without the specific association that the PRMD was related to playing an instrument. A broad variety of additional terms were applied, sometimes only in one or a few studies (see Tables 1 through 5).

Case–control study

The included case–control study (Sakai and Shimawaki 2010) investigated the indices of hand and movement angles in 220 pianists in Japan with overuse disorders and 62 unaffected pianists as controls. The authors reported that epicondylitis, muscle pain in the forearm and hypothenar region, De Quervain’s tendinitis and distal tendinitis were correlated with variable parameters of small hand size (Table 1).

Quality assessment of the case–control study

The quality of the case–control study (Sakai and Shimawaki 2010) was assessed as a − 3 (out of 14 possible points) due to relevant concerns in the study methodology, such as the absence of a sample size justification and not reporting the details of the study population, inclusion- or exclusion criteria or participant selection (randomly or as a convenience sample). Furthermore, there was no reporting on the blinding of the assessors of exposure/risk factors.

Cohort studies

Three of the six cohort studies were retrospective evaluations of patient records from a university’s health service, one of which was a follow-up of another study (Manchester and Flieder 1991; Manchester 1988; Manchester and Lustik 1989). The incidence of playing-related disorders of the upper limb and/or hand was reported as 8.5 episodes per 100 music students per year, and 16% had persistent complaints. Risk factors were not evaluated. The fourth cohort study investigated the incidence of work-related musculoskeletal disorders in music teachers and did not identify an elevated arm position of > 30° as a potential risk factor (Fjellman-Wiklund and Sundelin 1998). Another cohort study found that there is generally a moderate degree of disability and pain in violinists, cellists and pianists, though cellists had more disability and pain than the other types of musicians (Piatkowska et al. 2016). The sixth cohort study found that 29% of music students complain about playing-related health issues (mixed somatic and psychological) in their 1st year of study (Nusseck et al. 2017). Details are shown in Table 2.

Table 2 Characteristics and results of the included studies, cohort studies
Full size table

Quality assessment of the cohort studies

The quality assessment scores of the six cohort studies were between − 11 and 10 out of 16 possible points, mainly due to significant methodological concerns in most of the studies. Only three of the studies (Fjellman-Wiklund and Sundelin 1998; Nusseck et al. 2017; Piatkowska et al. 2016) had clearly defined study populations. None of the studies reported sample size calculations, and only one study (Nusseck et al. 2017) prospectively measured risk factors. However, three studies applied validated outcome measurements (Nusseck et al. 2017; Piatkowska et al. 2016; Kuorinka et al. 1987; Fjellman-Wiklund and Sundelin 1998).

Cross-sectional studies

We found 62 cross-sectional studies (see Table 3).

Table 3 Characteristics and results of the included studies, cross-sectional studies
Full size table

Outcome measures used in cross-sectional studies

MCDs in the cross-sectional studies were measured with different instruments and used different definitions and questionnaires that to our knowledge, were often not validated (Abréu-Ramos and Micheo 2007; Ackermann et al. 2011; Arnason et al. 2014; Blackie et al. 1999; Chan et al. 2013; Crnivec 2004; Davies and Mangion 2002; De Smet et al. 1998; Fishbein et al. 1988; Fry 1986; Heredia et al. 2014; Hodapp et al. 2009; Kim et al. 2012; Kok et al. 2013; 2015; Kovero and Könönen 1995; Marques et al. 2003; Schäcke et al. 1986; Middlestadt and Fishbein 1989; Molsberger 1991; Papandreou and Vervainioti 2010; Raymond et al. 2012; Steinmetz and Möller 2007; Shields and Dockrell 2000; Sakai 2002).

Period of time investigated in cross-sectional studies

The time period of the measured MCD varied substantially between studies, ranging from whole lifetimes, years playing an instrument, the previous 1, 3, 12 or 18 months, the last 7 days or current complaints. For some publications, we could not find a specification of the exact time frame used by the study, f.e. (Lima et al. 2015; Blackie et al. 1999; Crnivec 2004; De Smet et al. 1998; Eller et al. 1992; Fishbein et al. 1988; Fry 1986; Schäcke et al. 1986; Hodapp et al. 2009; Kim et al. 2012; Kovero and Könönen 1995; Middlestadt 1990; Middlestadt and Fishbein 1988, 1989; Molsberger 1991; Papandreou and Vervainioti 2010).

Prevalence reported in cross-sectional studies

Due to substantial heterogeneity among the measured complaints, a comparison of prevalence and correlations to the MCD was not feasible. In the included studies, the prevalence of MCDs ranged from point prevalence 0% in one small study (Logue et al. 2005) to 12-month prevalence 97% in a study of female orchestra musicians (Paarup et al. 2011). However, within the four studies that received a quality assessment of 5, the 12-month prevalence ranged from 82 to 86% (Leaver et al. 2011; Kaufman-Cohen and Ratzon 2011; Fjellmann-Wiklund et al. 2003), with 26% of musicians reporting current complaints (Nyman et al. 2007).

Risk factors reported in cross-sectional studies

The four studies with the best quality assessments reported playing with an elevated arm position as a risk factor for neck-shoulder pain (Nyman et al. 2007). Neck-shoulder discomfort in female music teachers was correlated with high psychological demands and teaching at multiple schools, whereas in male music teachers, it was associated with lifting, playing the guitar and low social support (Fjellmann-Wiklund et al. 2003). In symphony orchestras, MCD tended to be more frequent among women, in musicians experiencing low mood and in those with high somatizing scores. Only weak associations were observed with psychosocial work stressors and performance anxiety (Leaver et al. 2011). In musicians who play string instruments, the odds of wrist/hand pain were 2.9-fold higher than for those who play wind instruments (Leaver et al. 2011). In contrast, another study found that in classical musicians, string musicians showed higher PRMD scores than woodwind and brass players. Furthermore, the study found a correlation with biomechanical risk factors, perceived physical environment risk factors, instrument weight and average number of hours played per week (Kaufman-Cohen and Ratzon 2011). However, cross-sectional studies have low validity for verifying risk factors such as exposure time, and the occurrence of outcomes cannot thus be properly measured.

Quality assessment of cross-sectional studies

The quality scores of the 62 cross-sectional studies were between − 13 and 5 out of 16 possible points, with significant methodological concerns existing in the studies. Only one study (Nusseck et al. 2017) utilized a time frame that could reliably measure the association between the exposure and outcome, although several studies explicitly looked for risk or predicting factors. No studies reported the blinding of outcome measures. Sample size calculations were reported in only two of the studies (Kochem and Silva 2017; Kaufman-Cohen and Ratzon 2011). Frequent concerns included the objectivity, reliability and validity of the outcome and exposure measurement tools. Often, no confounders were assessed, and only a few studies assessed at least some important confounders (Eller et al. 1992; Fjellmann-Wiklund et al. 2003; Leaver et al. 2011; Nyman et al. 2007; Kok et al. 2015; Nusseck et al. 2017; Piatkowska et al. 2016; Woldendorp et al. 2016, 2017).

Intervention studies

Study designs and methods of the interventional studies included

Three of the 12 intervention studies were performed using pre–post study designs without a control group (Table 4). One of these three studies compared two different interventions, (strength vs. endurance training) stratified by instrument played, but further information on randomization was not provided (Ackermann et al. 2002b). The other nine interventions used controlled designs (Table 5). Seven of the 12 studies were randomized controlled trials. In one partially blind study, playing under the intervention or control condition was randomized (Ackermann et al. 2002a). In one study with a three-armed design, participants in the two intervention groups were randomized, but the control group was recruited separately (Khalsa et al. 2009). In one non-randomized study, allocation of the nine eligible orchestras followed geographical criteria; the six orchestras that were geographically closest to each other were selected as interventional orchestras, while the remaining three served as control orchestras (Brandfonbrener 1997). The second non-randomized study was a pilot study for a partially randomized trial that was conducted later. The pilot recruited its control group separately due to a low number of participants (Khalsa and Cope 2006; Khalsa et al. 2009). Two publications about Tuina treatment appeared to be drawn from the same study population, reporting immediate effects and effects occurring after 3 weeks (Sousa et al. 2015a, b).

Table 4 Characteristics and results of the included studies, pre–post studies without control groups
Full size table
Table 5 Characteristics and results of the included studies, controlled clinical trials
Full size table

Duration of study interventions

The duration of the mostly physiotherapeutic interventions ranged from a single application given on 1 day with directly measured effects to interventions lasting 3, 5, 6, 8, 9, 12 and 15 weeks or interventions lasting up to 1 year. In one study, the treatment duration was individualized to each participant; the median treatment duration was 27 months, with a standard deviation of 16.7 months (Steinmetz et al. 2009).

Study population of interventional studies

Five studies explicitly included participants with complaints to investigate the effectiveness of the interventions (Sousa et al. 2015a, b; Damian and Zalpour 2011; de Greef et al. 2003; Steinmetz et al. 2009). Six studies included both musicians with complaints and musicians without complaints and investigated the prevalence and intensity of complaints before and after the interventions (Nygaard Andersen et al. 2017; Ackermann et al. 2002b; Brandfonbrener 1997; Chan et al. 2014; Khalsa and Cope 2006; Khalsa et al. 2009) or during the interventions compared to control conditions (Ackermann et al. 2002b).

Used outcome parameters of interventional studies

The intervention studies did not reveal evidence that the interventions were effective. No statistically significant changes in MCD were found in violinists after scapula taping, in musicians participating in yoga lifestyle interventions, or in orchestra musicians attending music didactic lectures and receiving instructions for home exercises or trigger point treatment with radial shockwave therapy combined with physical therapy or specific strength training in comparison to a control group (Ackermann et al. 2002a; Khalsa and Cope 2006; Khalsa et al. 2009; Brandfonbrener 1997; Damian and Zalpour 2011; Nygaard Andersen et al. 2017). However, two studies reported significant improvements within the treatment groups but not in the control group (Damian and Zalpour 2011; Nygaard Andersen et al. 2017). Tuina application and groningen exercise therapy did produce statistically significant reductions in pain in the treatment groups compared to the control groups (de Greef et al. 2003; Sousa et al. 2015a, b). Details are provided in Tables 4 and 5.

Table 6 Characteristics and results of the included studies, case studies
Full size table

Quality assessment of interventional studies

Pre–post studies without control groups The quality assessment scores of the three studies in this category were rated between − 3 and 4 out of 15 possible points. Overall, there were significant methodological concerns. None of the studies reported sample size calculations, sufficient blinding or the use of objective, reliable and validated outcome measures. Inclusion and exclusion criteria were explicitly stated in only two studies (Ackermann et al. 2002b; Steinmetz et al. 2009). In one study, a group of participants received physiotherapy in addition to the examined intervention, but a subgroup analysis was not performed (Steinmetz et al. 2009).

Controlled clinical trials The quality assessment scores of the nine studies in this category ranged between − 10 and 7 out of 18 possible points. Overall, noticeable methodological concerns included not reporting on how blinding was conducted, not using reliable randomization procedures and not providing sample size calculations. Only one study (Ackermann and Adams 2003) reported an analysis or avoidance of additional treatments. Only a few studies reported an intention-to-treat-analysis or reported using all the participants in their assigned groups for the analyses (Sousa et al. 2015a, b; de Greef et al. 2003; Nygaard Andersen et al. 2017). In four studies, the outcome measurements were not considered to be objective, reliable and validated (Brandfonbrener 1997; de Greef et al. 2003; Khalsa et al. 2009; Khalsa and Cope 2006).

Case studies

There were 28 case studies included, most of which were retrospective single case reports (Table 6). Almost all case studies described improvements in the complaints following the respective treatments or associations between risk factors and special complaints. We were unable to identify subsequent intervention studies to verify the treatment effects. The reported effects included surgical decompression procedures in nerve compression syndromes (Hoppmann 1997; Laha et al. 1978; Miliam and Basse 2009), surgical treatment in neurogenic thoracic outlet syndrome (Demaree et al. 2017), the use of individual orthotic or assistive devices (Anderson 1990; Dommerholt 2010; Price and Watson 2011; Sakai 1992; Wilson 1989), conservative combined treatments (Lederman 1996; Patrone et al. 1989) partly involving posture optimization (Manal et al. 2008; McFarland and Curl 1998; Potter and Jones 1993, 1995; Quarrier and Norris 2001) and physiotherapy with a special emphasis on mobilization of the radial and posterior interosseous nerves (Jepsen 2014), tensegrity massage and advice for positioning or leisure activities (Wilk et al. 2016). One report each exists for acupuncture (Molsberger and Molsberger 2012), EMG biofeedback (Levee et al. 1976), myofascial trigger point therapy (Dommerholt 2010) and the Feldenkrais method (Nelson 1989).

Excluded studies

At the full-text level, 195 studies did not fulfill the inclusion criteria and were not included. Examples of the excluded studies are provided as follows. The study population included participants under 16 years of age (Vinci et al. 2015; Rodríguez-Romero et al. 2016; Stanek et al. 2017; Ioannou and Altenmüller 2015; Yasuda et al. 2016; Hagberg et al. 2005; Rodriguez-Lozano 2008; Goodman and Staz 1989; Larsson et al. 1993; Mehrparvar et al. 2012), did not state the age range of participants (Zaza and Farewell 1997; Lopez and Martinez 2013; Manchester and Park 1996; Kreutz et al. 2008; Revak 1989; Brandfonbrener 2000; Lee et al. 2012; Tubiana and Chamagne 1993; Williamon and Thompson 2006; Zetterberg et al. 1998) or provided insufficient information on the outcomes of interest when separated into age subgroups (Furuya et al. 2006; Dawson 2001). Further reasons for exclusion included that the professional status was not clearly specified (Pedrazzini et al. 2015; Takata et al. 2016; Mehrparvar et al. 2012; Brandfonbrener 2000, 2002; Tubiana and Chamagne 1993; Lederman 2003), MCDs were not reported discriminatingly (Hiner et al. 1987; Spahn et al. 2001) or complaints were not clearly separated into non-musculoskeletal regions in the reported results (Kaneko et al. 2005).

Discussion

The study designs, terminology, and outcomes of the 109 studies included in this review were heterogeneous. The inclusion criteria were rarely mentioned throughout all study types, the analyses mostly did not check for major confounders, and the definition of exposure was often insufficient. In addition, the quality assessment of most studies included raised considerable methodological concerns. Therefore, sufficient statements cannot be provided for the prevalence, risk factors, prevention and effectiveness of treatment of MCD in professional musicians. Furthermore, a more profound differentiation, for example with regard to gender or the instrument played, is even less feasible.

In musicians´ medicine, we regularly observe that professional musicians suffer from MCD due to high physical and psychological work-related demands. The wide range in the prevalence of MCD, reported in the included studies may reflect the heterogeneous and often unclear definitions used to assess complaints in musicians. We assume that causality was already inferred due to the use of terms such as “PRMD” or by asking the study participants about complaints that were caused by or noticed while playing music. Thus, the basic assumption of causality was not questioned. In musicians´ medicine, we often lack validated assessment tools for MCD in relation to making music. However, some tools do exist, such as the Disabilities of The Arm, Shoulder and Hand (DASH) questionnaire (SooHoo et al. 2002; Hudak et al. 1996) with its optional sports/performing arts module.

From an epidemiological viewpoint, most of the included studies had only a small number of participants. Only one cross-sectional study (Fishbein et al. 1988; Middlestadt 1990; Middlestadt and Fishbein 1988) included a high number of participants (n = 2212). who were recruited from various US orchestras. Only this study was considered representative. However, even this study had considerable methodological shortcomings (see above).

The included studies had mostly a high risk of bias and little control of confounders. Some cross-sectional studies have been conducted on musicians in a single orchestra during a single rehearsal session. These studies likely suffered from a healthy-worker bias in addition to having little external validity or representativeness in the study results. When biographic data are gathered in cross-sectional studies, recall bias likely exists. Very few of the studies systematically assessed confounders that may have influenced MCD independently of the treatment, such as other physical loads, increased physical activity, working in the house and garden, unilateral strain by special sport discipline, carrying toddlers or providing home care to relatives. Individual preventive activities, such as sports, physical training, and relaxation, were not included as possible confounders. Additionally, correlations between instrument-specific workload (e.g., daily playing time) and complaints were not assessed. To assess the risk factors for MCD in musicians’, observational cohort studies are needed that utilize a sufficient sample size, different instrument groups, workloads, or workplaces and account for physical activity and other risk factors. Clinical trials investigating conventional and complementary medicine approaches that differentiate between instrument groups are needed to assess effective therapy options for musicians.

Comparing our results to the literature we found seven systematic reviews about MCD in adult professional musicians (Zaza 1998; Bragge et al. 2006; Wu 2007; Baadjou et al. 2016; Jacukowicz 2016; Kok et al. 2016; Vervainioti and Alexopoulos 2015). The presented review is the first attempt to investigate the frequencies, risk factors and treatment options for MCD in musicians. Zaza (1998) reported that the prevalence of PRMD in adult classical musicians was comparable to that of work-related MCDs reported for other occupations. Due to the previously mentioned lack of evidence for the prevalence of PRMD, we are unable to draw a comparable conclusion. This finding is in line with another systematic review (Bragge et al. 2006) that concluded that the evidence did not provide sufficient information regarding the prevalence and risk factors associated with PRMDs in pianists due to common methodological limitations, including sampling/measurement biases, inadequate reporting of reliability/validity of the outcome measures, a lack of operational definitions for PRMD and a lack of statistical significance testing. In her work on the occupational risk factors of MCD in musicians, Wu (2007) reported that the etiology of MCD is multifactorial in instrumental musicians; however, she also pointed out that cross-sectional studies cannot investigate causality. This differs from our conclusion, since we could not identify studies that searched for risk factors using an appropriate study design. More recent systematic reviews also found methodological concerns (Baadjou et al. 2016; Jacukowicz 2016; Kok et al. 2016; Vervainioti and Alexopoulos 2015). In contrast to our review, the authors reported results from the subjects of each study. Baadjou et al. reported that previous musculoskeletal injuries, music performance anxiety, high levels of stress and being a female who plays a stringed instrument seem to be associated with more MCD (Baadjou et al. 2016). Kok et al. found a point prevalence of MCD in professional musicians ranging between 9 and 68%, a 12-month prevalence between 41 and 93% and a lifetime prevalence between 62 and 93%. Ten out of 12 studies show a higher prevalence of MCD among women (Kok et al. 2016). However, the authors used a different scoring system to assess the quality of the included studies. In contrast to our assessment, the authors found that 13 out of 17 studies received high-quality scores (Kok et al. 2016). On the other hand, the results of the present study agree with the authors who reported that the current definition of PRMD does not provide causality of the complaints. We further emphasize the importance of using adequate and validated instruments for measuring outcomes in future studies. Jacukowicz reported on the psychosocial aspects of work, such as long hours at work, work content, high job demands, low control/influence, and a lack of social support were related to MCD (Jacukowicz 2016). Another systematic review included professional musicians as well as active members of a classical orchestra who were of at least 16 years of age. The review proposed that further research should include seven categories of stressors that affect classical instrumental musicians: public exposure, personal hazards, repertoire, competition, job context, injury/illness, and criticism (Vervainioti and Alexopoulos 2015).

Limitations and strengths

The presented systematic review is a comprehensive attempt to evaluate the quality of the available literature on the prevalence, risk factors, and effectiveness of prevention or treatment of MCD in professional musicians. Therefore, we included published observational studies (case–control studies, cohort studies and cross-sectional studies), interventional studies (controlled clinical trials and pre–post intervention studies without control group), case reports and case series reporting clinical interventions. Because the data were very heterogeneous, an assessment of publication bias, such as a funnel plot, was not feasible. The quality assessments of the included studies were performed according to the study protocol, and if a methodological procedure was not described in the assessed article, it was recorded as not done. This was very often the case, with very few procedures being explicitly described as not having been performed. This may have led to poorer evaluations of those studies that provided less detailed reporting or publications that were older than the current reporting guidelines. The real study method might have been underestimated based on the shortcomings of reporting. The authors were not contacted about the included studies, because several of the studies were performed more than 10 years ago. Regarding quality assessment and assessment of the bias risk of the original studies, a certain allowance for interpretation and discussion is generally required. Therefore, quality evaluations were each carried out by two people, and a consensus was drawn for each study after a detailed discussion. A limitation of the present review is that to allow for a quantitative estimation, the quality assessment instruments, which were modified by the authors, and the subsequent quantitative estimation were not validated previously. The term MCD comprises a large number of very different disorders. The aim of this review was to provide an overview of this research area as well as to search for clues about the causality between music making and diseases or disorders. Any restrictions to specified clinical diagnoses would, therefore, have been purely arbitrary and, given the conceptual uncertainty of numerous primary studies with many-sided overlaps, would have represented a random and, therefore, meaningless section of a complex, interwoven area.

Conclusions

The body of evidence regarding musculoskeletal disorders and complaints in professional musicians has grown substantially, since the publication of earlier reviews. However, studies analyzing prevalence, risk factors and effectiveness of the prevention or treatment of MCD amongst professional musicians, using today’s methodological requirements, are still missing. To evaluate the extent of associations of practice and performance burden with MCD, prospective, long-term cohort studies that properly take into account the influencing factors are still needed. To evaluate the effectiveness of specific treatment options for specific instrumental groups, prospective randomized confirmatory intervention studies are necessary. The use of well-defined diagnostic criteria for MCD in musicians is needed to avoid bias when selecting study participants. Strict and consistent diagnostic criteria would help to avoid the large variation in results. Establishing and implementing a validated and reliable set of outcome measurements is necessary. If it is possible to optimize the methodology as proposed above, relevant risk factors for MCD in musicians may be identified more precisely and allow for targeted prevention and intervention.