Musculoskeletal Injuries in Dance: A Systematic Review

N=239 Articles rejected on basis of title and abstract failure to comply with retrieval criteria: The study was based on ballet or any forms of artistic dance and had as its focus musculoskeletal injuries or pain, or screening for injury prevention or interventions to reduce musculoskeletal injury or pain.


Introduction
Sport participation entails a risk of injury. Part of the responsibility of those charged with caring for sports persons is to mitigate that risk. Some of this may be achieved through the introduction of measures to minimise injuries. Dance participation, like sport, can lead to injury, however, unlike most and elite sports persons, many dancers do not have ready access to specialist or in-house medical teams. With the absence of international (e.g. International Olympic Committee, Fédération Internationale de Football Association-FIFA, and International Rugby Board) or national legislative governing bodies (e.g. Football Association, Rugby Football Union) dance has a relative lack of accountability governing how their participants are supported. As part of an injury prevention strategy in dance, an understanding of the extent of injuries is needed. Furthermore, knowledge of the impact of specialist and in-house medical provision in reducing injuries is required to inform those involved in delivering dance training and performance.
To date two systematic reviews pertaining to musculoskeletal injuries in dancers have been published. The first was designed to assemble and synthesize, using the "best evidence synthesis" approach, the epidemiology, diagnosis, prognosis, treatment and prevention of musculoskeletal injuries and pain in the dancing population up to October 2004 [1]. Through the application of a priori criteria 32 articles were accepted for review. The authors comment that 69% of the articles identified from the titles and abstracts were consequently not accepted following full text review due to lack of scientific vigour. Of the studies accepted Hincapie et al. [1] indicates the literature has many limitations resulting in difficulty in drawing consistent conclusions. The limitations include: the variety of injury definitions used; the heterogeneous nature of the populations; failure to identify the population at risk that should form the denominator in incidence (or prevalence) calculations; and the wide range of inclusion and exclusion criteria. Hincapie et. al. [1] does offer some important conclusions despite these limitations, including evidence that musculoskeletal injury is an important issue for all dancers and that there is preliminary evidence that comprehensive injury prevention and management strategies may reduce injuries. In a follow-up to the original Hincapie et al. review [1], Jacobs et al. [2] extends the review period from October 2004 to March 2008. Repeating their methodology the authors reviewed a further 19 articles that were deemed scientifically admissible (with similar reasons for exclusion noted as per their previous review). The authors comment on an increase in the percentage of admissible studies rising to 68% compared to the 31% noted in the previous review. Jacobs et al. [2] still reiterate the need for explicit criteria on injury definition and methods of injury reporting and comment that there are still major scientific limitations and biases in the literature reviewed.
These two systematic reviews demonstrate that it is important to evaluate the evidence on musculoskeletal injuries in dancers and that due to an increasing emphasis on scientific rigor in dance medicine related articles, an up-to-date position needs to be established. Guidelines based on the evidence needs to be established that acknowledge the evidence profile from which they are based. There is a need to create recommendations for dance using recognised methodologies such as the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) system, to evaluate both high and low level evidence [3][4][5][6][7][8][9][10][11][12][13][14].
The objectives of this review was to undertake an up-to-date systematic review and evaluation the level of evidence of the literature pertaining to musculoskeletal injury in dancers using the GRADE system. Furthermore to examine the impact that specialist dance medicine provision, including in-house medical teams, screening and exercise programmes, has on overall dance injury incidence and using the GRADE system to establish the level of evidence and strength of recommendations for its use in reducing the overall incidence of injuries in dancers.

Method
This systematic review of literature pertaining to musculoskeletal injuries in dance and evaluation of the impact of specialist dance Potential relevant articles identified and screened using the following MeSH terms: Dance AND: injury, injuries, epidemiology, rehabilitation, treatment, prognosis, diagnosis, pain, prevention, screening, musculoskeletal, incidence, prevalence AND Ballet AND: injury, injuries, epidemiology, rehabilitation, treatment, prognosis, diagnosis, pain, prevention, screening, musculoskeletal, incidence, prevalence N=3055 Articles retrieved for in-depth review due to compliance to retrieval criteria or inability to determine inclusion based on title or abstract. N=239 Articles rejected on basis of title and abstract failure to comply with retrieval criteria: The study was based on ballet or any forms of artistic dance and had as its focus musculoskeletal injuries or pain, or screening for injury prevention or interventions to reduce musculoskeletal injury or pain. N=2816 Articles rejected based on exclusion criteria: recreational, social or aerobic dance forms; single case or single pathology studies; psychological aspects as the primary focus; clinical evaluations of treatments for a named/specific pathology; non-English language studies, chapters in books, abstracts and poster presentations. N=192 Articles accepted for Systematic Review based on inclusion criteria: The study was based on ballet or any forms of artistic dance and had as its focus musculoskeletal injuries or pain, or screening for injury prevention or interventions to reduce musculoskeletal injury or pain. N=47 medicine provision on overall dance injury incidence was undertaken using the GRADE system. The value of implementing GRADE is the ability to evaluate high and low level evidence studies. This is an important aspect when considering the evaluation of dance injury literature as it has already been established to be of a low level. Furthermore the value of GRADE is that the direction and strength of recommendations takes into consideration the level of evidence from which it is drawn but evaluates if the proposed benefits outweigh any harm and so may differ from the evidence profile. The use of the Evidence to Recommendation Framework enhances the transparency of those recommendations. The GRADE system also allows studies with, for example differing injury definitions to be included to improve the sample size from which the data is drawn, but downgrades the overall confidence in results due to its evaluation of its heterogeneous nature [3][4][5][6][7][8][9][10][11][12][13][14].

Literature Review
A framing question was set prior to commencing the literature review: To evaluate the available literature from 1966 to 2013 to determine the level of evidence around musculoskeletal injury rates in dancers and the potential impact that specialist dance medicine provision may have on overall injury rate and pain. For the purpose of inclusion, "specialist dance medicine provision" programmes included implementation of in-house dance medicine teams, screening and exercise programmes [15][16][17][18][19][20][21][22][23][24][25][26][27].
A systematic search of the scientific literature was then undertaken using the following electronic databases: the Cochrane Library; Medline . If there was insufficient information in the title or abstract to determine its inclusion a full text manuscript was retrieved and a review was undertaken with the full inclusion and exclusion criteria applied. Applying the inclusion and exclusion criteria (Figure 1), two authors appraised the relevance of each identified study in order to agree the final list of included studies [28][29][30][31][32][33][34][35][36][37][38][39][40][41][42][43][44].

Critical Review of Included Literature
Each included study was summarised in a Table 1. This included data relating to methods, sample size and duration, injury definition, incidence or prevalence outcomes, severity outcomes and nature of intervention. The data collected allowed two key patient important outcomes to be determined, namely Injury Rate as an important outcome and Injury Reduction as a critical outcome. These outcomes were then used to determine the quality of the evidence presented and the strength of subsequent recommendations [14].
The rating of evidence was achieved through the Evidence Profiles for the respective patient important outcomes within which an evaluation of the evidence in relation to limitations, inconsistency, indirectness, imprecision and publication bias across the studies. Optimal Information Size (OIS) was calculated using a sample size calculator: the OpenEpi sample size calculator for a descriptive study where sample size: (http://www.openepi.com/OE2.3/SampleSize/SSPropor.htm). The Evidence Profile for injury incidence includes a calculation of the mean injury incidence per 1000hrs with 95% Confidence Intervals (CI) and number of injuries per dancer per year across the relevant studies with 95% CI. The Evidence Profile for injury reduction includes a calculation for the pre-intervention mean incidence/1000hrs, range and 95% CI, or number of injuries (and range) and a test/post-intervention mean injury incidence/1000hrs or number (and range). This repeated measures methodology is employed due to the absence of any control groups used in the intervention studies.
The development of the recommendations for the overall reduction of injuries in dance involved the balance between desirable and undesirable outcomes and the application of patient focussed values and preferences to determine the direction of the recommendation. These issues were taken alongside the quality of the evidence to then determine the strength of the recommendation.

Results
A total of 3055 titles and abstracts of studies were reviewed. Applying the retrieval criteria 239 studies were retrieved for full, indepth review. Using the inclusion and exclusion criteria, 47 were then accepted and preceded to evaluation as part of this systematic review ( Figure 1).
The majority of the studies related to ballet and modern or contemporary dance but other styles included break-dance, hip hop, theatrical dance, Mexican, Spanish, tap, Morris, flamenco, Irish, highland, jazz or a mix of styles. Participants were from professional or competitive dance backgrounds as well as student or vocational dance environments. Of studies accepted, 36 were conducted using a retrospective methodology including review of medical records or surveys while only 9 were prospective methodologies with a further two studies combining both retrospective and prospective components. Injury definitions varied between time loss to medical records or seeking medical attention to financial outlay. Some studies did not explicitly indicate the injury definition used. Details of injury were reported in a number of ways, including incidence, prevalence, injury per dancer or total numbers of injuries but very few papers reported severity of injuries (Table 1).
Within the Evidence Profile for Injury Rate (Table 2) a mean incidence of 1.33/1000hrs for the 12 observational studies that reported incidences of injury was calculated. An average of 1.93 injuries per dancer per year was calculated from 29 studies that had sufficient data.
These studies demonstrated serious limitations (including the lack of the inclusion of control populations and flawed exposure measurements), and inconsistencies (including the discrepancy of results within the studies and heterogeneity). There was no imprecision (on account of the sample size (n=2788 and 5318) being greater than the sample size calculated (n>385) [61]), indirectness, or publication bias noted. This resulted in a downgrading of the Low evidence assigned to observational studies within the GRADE system to Very Low.
The Evidence Profile for Injury Reduction noted a reduction from a mean incidence of 2.46/1000hrs to 0.84/1000hrs. There are no comparison groups for these studies as repeated measures methodologies were employed across the entire cohort studied. These 2 observational studies demonstrate serious limitations (due to the lack of the inclusion of control populations), inconsistency (due to the range of results reported and heterogeneity) and imprecision (due to the sample size n=363 being less than the sample size calculated n>385). Publication bias is also likely with both studies demonstrating significant reductions in injury incidence and thus more likely to be favoured for publication over studies that failed to demonstrate statistically significant findings. With these factors in mind the Evidence Profile rating is downgraded from Low for observational studies to Very Low (Table 3). A further 3 observational studies demonstrated a reduction in injury numbers from 137 to 106. These studies also demonstrated serious limitations (due to the lack of the inclusion of control populations) and imprecision (due to the less than optimal information sample size) and so the Evidence Profile was rated down from Low to Very Low (Table 4).
Using the GRADE framework for moving from evidence to recommendation, a recommendation for the use of comprehensive medical management for the reduction of injury rate in dancers is advocated in the absence of stronger evidence (Table 5).

Discussion
The aim of this review was to provide an up-to-date systematic review of the literature pertaining to musculoskeletal injury and pain in dancers using the GRADE system to establish the level of evidence and strength of recommendations for reducing the overall incidence of injuries in dancers. When examining the literature retrieved through this systematic review similar findings to the two earlier studies are noted with many limitations including the variety of injury definitions used, the heterogeneous populations; identifying the at risk population and the wide range of inclusion and exclusion criteria [1,2]. An overall injury incidence of 1.33/1000hours, and an injury reduction as a consequence of comprehensive medical management from 2.46/1000hrs to 0.84/1000hours was calculated.

GRADE evidence profile
One of the fundamental aspects of the GRADE system is that sequential judgements are made regarding the quality of evidence across studies for each patient important outcome [4]. It determines which outcomes are critical to a decision and the overall quality of evidence across those critical outcomes. This includes the balance    between benefits and harms, and the strength of the subsequent recommendations [62].
From the literature retrieved, 29 studies were used in an evidence profile for Injury rate (injuries/dancer/year) of which 12 allowed injury incidence to be considered. With the studies there were serious limitations or biases noted due to the failure to include control groups, flawed measurements of exposure and failure to control confounding variables including a failure of accurate measurement of all known prognostic factors and failure to match for prognostic factors. The studies in question were also noted for serious inconsistency due to the heterogeneous nature of the contributing studies. It is well appreciated that various dance styles provide differing challenges on the body [63,64] that may result in injuries as well as the potential differences in injury potential noted between professional and student participants [15,16,26]. As the pooled sample size for Injury Rate (n=2788 and n=5319) was greater than the n=385 generated by a conventional sample size calculation for a single adequately powered trial it was not rated down for imprecision. There was no evidence of publication bias noted in this outcome group. As a consequence of these issues in the absence of any upward rating through magnitude of effect, dose response or confounders likely to minimise the effect, the overall rating of Low evidence for observational studies is downgraded to Very Low as a reflection of the overall confidence in the effects.
Similar issues over evidence were observed when considering Injury Reduction as a patient important outcome. There were serious limitations or biases noted due to the failure to include control groups, flawed measurements of exposure and failure to control confounding factors. The studies in question were also noted for serious inconsistency due to the heterogeneous nature of the contributing studies, although these studies were limited to professional ballet and modern dancers as opposed to the range of patient groups noted in the Injury Rate Evidence Profile. The pooled sample size (n=363) for Injury Reduction was less than the required sample size of 385 so therefore was rated down for imprecision. It was also noted that publication bias was likely due to the statistical significance reported in studies resulting in their acceptance for publication as opposed to studies that may not have demonstrated significant findings. As a consequence of these issues the overall rating of Low evidence for observational studies is downgraded to Very Low.

GRADE recommendations
The nature of the important outcomes were decided from the perspective of the patient as opposed to the funders of dance related healthcare systems, thereby putting greater emphasis on reduction of injuries as opposed to costs of service/resource. Although rated as very low evidence, suggesting that the true effect may be much larger or smaller, there is sufficient call to consider means to reduce the overall injury rate in dancers. The role of comprehensive medical management as a means to address the important patient outcome of Injury Reduction was demonstrated to have a Very Low evidence profile. Using the GRADE framework a strong recommendation for the use of comprehensive medical management for the reduction of injury rate in dancers is advocated in the absence of stronger evidence ( Table 5).

No of studies (design) limitations inconsistency indirectness Imprecision Publication bias
Control/ pre-intervention: Average injury numbers (Range of total injury numbers)  x We can be confident that professional dancers place a high value on a reduction in injury rate as their livelihood is dependent on their ability to dance The increasing number of higher quality studies into injury rate reflects the position of the dance environment Resource implications: Are the resources worth the expected net benefit from following the recommendation?
x There is a resource need to provide in-house medical provision. This has been demonstrated to reduce the overall medical costs and outweigh the costs of its implementation.
Although not explicitly examined as part of the review, the use of in-house medical teams are becoming more common place-the implementation of injury audits, screening and program interventions could be seen as sunk costs. Cost per resource unit needs to be established.
Overall strength of recommendation Strong The author recommends that the injury rate of dancers in professional companies will be reduced through the use of comprehensive medical management.
Evidence to recommendation synthesis The high value placed on injury reduction through comprehensive medical management versus harm outweighed the lower evidence profile in the absence of stronger evidence The value to this system is that it allows for a strong recommendation to be made despite a lower level of evidence presented. This is through basing the recommendation on patient important outcomes and evaluating the benefit versus harm or that the desirable effects outweigh the undesirable effects in respect to the intervention. The use of comprehensive medical management for professional athletes reflects more a duty of care in modern sports medicine and so fits with a strong recommendation for its implementation in dance in the absence of higher evidence. This study also helps to identify the low level of evidence available supporting the incidence and intervention for dance and dance injuries. Previous international consensus statements in sport have been agreed to improve the methodology for the collection of injury data [65][66][67]. Within these consensus statements, there is an advocacy towards methodologies that utilise prospective cohort studies with time loss injury definitions and calculated exposure, allowing incidence to be calculated as injuries per 1000hours of sports participation. The process of reaching (and using) consensus definitions has also been recognised in dance [68][69][70]. There is a need to consider a similar consensus statement for injury data collection in dance to improve methodological rigor and consistency between studies in line with those recommendations in sport.

Limitations
Although this systematic review was conducted using Medline based electronic databases, specialised journals and grey literature, the exclusion of a full PubMed due to the requirement of medical subheadings in the search process, may result in some relevant studies not being identified. The exclusion of unpublished work/thesis, poster presentations and abstracts along with chapters from books may also reduce the total number of studies available from which evidence and recommendations can be drawn. Similarity the exclusion of non-English language studies is a further limitation.

Conclusion
The two previous systematic reviews concluded that the quality of evidence surround musculoskeletal injury and pain in dancers was low. The results of this systematic review were similar when using the GRADE system. Using the GRADE system two patient important outcomes, namely injury rate and injury reduction, were examined across the studies retrieved and an overall rating of evidence for both outcomes was very low. The value of implementing GRADE is the direction and strength of recommendations may differ from the evidence profile if the proposed benefits outweigh any harm. The use of the Evidence to Recommendation Framework enhances the transparency of those recommendations. A recommendation for the use of comprehensive medical management for the reduction of injury rate in dancers is advocated in the absence of stronger evidence. In the absence of an authoritative governing body, we would hope that those organisations involved in training or performances involving dancers would consider this recommendation. In an attempt to improve the methodological rigor in dance injury studies we advocate the development of a consensus statement along the lines achieved in other sports.