Injury Literature Critiques
by Ya-Ti Lin


Lists of Literatures

Other Literatures

Articular Mobility in Ballet Dancers: A follow-up study after four years

My Critique on: Patrick Klemp, FCP, Derek Chalton. Articular Mobility in Ballet Dancers: A follow-up study after four years. American Journal of Sports Medicine  1989;17:72-75.

Purpose and Objective
Due to the stigma that ballet dancers should be extremely flexible or even hypermobile, this study shows an association between hypermobility and the continuation of dancing, but being hypermobile is not a necessary factor to excel as a dancer. Also, the study shows that hypermobility is not more prevalent in ballet dancers than in non-dancers. This is an interesting topic in the field specifically target the general ballet dancers around the world. Unfortunately, the study designs are not proper enough in order to respond to the questions the authors intend to explore.
Design of the Study
The authors use multiple study designs to answer several different hypotheses. First, the one-group design is conducted, and thus, each dancer serves as her own control. The Beighton・s modification of the Carter and Wilkinson method of assessing articular hypermobility is used as criteria to ascertain each dancer with score from 0 to 9. These scores of 377 ballet dancers from the University of Cape Town Ballet School and the Cape Performing Arts Board (CAPAB) Ballet Company were obtained initially in the year of 1981. After excluding 150 dancers: 47 professional dancers due to their dance training is a lot longer than the student-dancers and the other 103 dancers couldn・t be traced after 4 years, the lost to follow-up of the 103 dancers could be a source of selection bias in interpreting the later analyses.
Of the 227 dancers who are ascertained to be included in the study, after 4-year follow-up, 55 who remain at school and the remaining 172 dancers, almost 76% of the source population, left the school due to the reasons, such as dancing elsewhere, non-progression due to lack of dancing ability, becoming teachers, injuries, loss of interest in dancing, and psychological, domestic, or financial reasons. Though the second examination scores of all 227 were known, the groups of dancers who had left the ballet school prior to the 4-year period can be distinctly different from the group remains at school.
There are several cross-sectional analyses performed in the article to observe different associations of the duration of training and acquired forward flexion, hypermobility and continuation of dancing, and the articular mobility score and dance achievement.
The diagnostic criteria of score of the Beighton・s method can be subject to the observer・s subjectivity even though pre- and post-examination are evaluated by the same person. Non-random errors can occur due to the observer・s mental, physical, and environmental conditions. Furthermore, the problem of the definition of hypermobility as the score greater than or equal to 4 is problematic and the subjects of scores 3 or 4 are more likely to be misclassified and the result can be misinterpreted. The selection of dancers and the large number of exclusion due to unknown reason makes the whole study vulnerable to selection bias and its external invalidity.
Presentation of Findings and Data Analyses
The Beighton・s scale is ordinal; therefore, the paired t test couldn・t be used in the significance testing. It is more appropriate to use the nonparametric method in this study because the sample size is small and the assumptions about the base shape of the distribution are not well established and the central-limit theorem doesn・t seem applicable here. The Wilcoxon matched-pairs test is performed in the some the data analyses.
In Table 1, a cross-sectional analysis of the association of the ability to forward flex and the duration of dance training is drown by the authors to conclude that because forward flexion is largely acquired through training, it should not be used as a criterion for assessing articular mobility in ballet dancers. It is noted that the comparability of the 50 dancers, with 29 had been able to forward flex as cases when examined initially and 21 dancers as non-cases, is questionable and the authors do not mention about their controlling for confounding factors such as sex.
In table 2, a cross-sectional analysis to come up with the positive association between hypermobility and the continuation of dancing, the authors gathering 55 dancers at school and 47 others who dance elsewhere to represent the continue dancing population in general and compare these 102 dancers with the 70 dancers who have stopped dancing after 4-year period could pose some problems on their selections of cases and non-cases and also their representation of the population to which the authors want to extrapolate. The evaluation of the arbitrary 4-year duration can be misleading and further information within the 4 years is needed to observe the trend.
Another cross-sectional analysis is performed to assess the hypermobility between the 47 professional ballet dancers and the controls matched for age and sex. It is extremely unclear that how and where the controls of the nondancers are selected since the authors don・t mention it in the text. By the meager information of 2 of 47 (4.3%) professional dancers are hypermobile versus 1 of 47 nondancers is hypermobile, it is risky to generalize the conclusion to all dancers because the representation of the 47 professional dancers in the CAPAB Ballet Company in South African can be very different from the professional dancers in the American Ballet Theatre or the New York City Ballet, for example. Therefore, it is difficult for the readers to tell whether is it justifiable to agree with the authors・ conclusion that hypermobility is not more prevalent in ballet dancers than in nondancers. The same external validity problem applies to another conclusion by the authors that there does not appear to be an association between hypermobility and dancing excellence.
Lastly, but not least, the study does not answer the authors previously stated question, :After how many years of dance training is forward flexion acquired?; If the authors want to assess the cause-effect association between the duration of ballet training to dancers・ acquirement of articular mobility, the study designed by the authors is a 4-year follow-up period for the same individuals to be reexamed, and this becomes irrelevant to respond to the question of :how many years".
The authors conclude that most ballet dancers can forward flex, that in the majority of cases forward flexion is acquired through training and occurs when exercises to increase suppleness are introduced into the training program, and that while there appears to be an association between hypermobility and the continuation of dancing, it is not necessarily the hypermobile individual who excels as a dancer. Also, hypermobility is neither common in ballet dancers nor is it necessarily an advantage for a dancer to be hypermobile in order to have a successful career. The problems of concluding the above statements are discussed earlier in this critique.
Classical ballet is a demanding professional occupation, with participants who are often underserved in terms of accurate diagnosis and appropriate comprehensive medical care. Largely of this stigmatization is due to lack of epidemiologic research and lack of data existing for easy and immediate access. The difficulty of conducting an epidemiologic research in ballet injuries is perceivable due to the limitation of resource and the idiosyncratic population the dancers are. The Hazards for ballet dancers to get injured occupationally can be personal, economical, psychological, and physical. They have the constant pressure and fear of being laid off and they, often times, conceal their injuries. Apart from their fear of unemployment, other barriers for ballet dancers for dancers to receive treatment are misunderstanding from the healthcare community, cost of treatment, time constraints, and behaviorally about dancers・ misperception of injuries and pains are normal and they have no choice but live with them.
It is hope that in the years to come, more researches be funded and conducted specifically target the ballet dancer population in the field of injury epidemiology to allow this underserved population be more visible and the ultimate goal is to make the prevalence and the incidence of the ballet injuries down to minimum possible.

An Analysis of Race and Demographic Factors among Motor Vehicle Fatalities

Jason Haukoos and I collaborate on the critique: Mayrose J., Jehle D.V.K. An Analysis of Race and Demographic Factors among Motor Vehicle Fatalities. The Journal of Trauma Injury, Infection, and Critical Care 2002;52:752-755.
Motor vehicle collisions account for significant morbidity and mortality in the United States, and if used properly, safety belts may prevent such morbidity and mortality. Although, safety belt use has increased over the past several decades, only approximately 70% of all passengers in motor vehicles regularly utilize restraint devices.
This study was performed to determine if demographic characteristics were associated with safety belt use among motor vehicle occupant fatalities. Although not specifically stated, it was the authors?hypothesis that differences existed among different demographic categories. The intent of this study was to identify, with the goal of targeting, a group or groups of people who do not regularly wear safety belts, and therefore are at high risk for injury or death.
Study Design
This was a retrospective cohort study that used data from the Fatality Analysis Reporting System (FARS) created by the Department of Transportation and the National Highway and Traffic Safety Administration (NHTSA) between 1993 and 1995. Demographic data, including age, sex, race, position in the vehicle, and whether a safety belt was used, were obtained for fatal occupants of motor vehicle collisions. Race is not included in the FARS database and was, therefore, obtained from linked death certificates. Inclusion criteria were not specified, although they are detailed elsewhere(1) and included data gathered from police accident reports, death certificates, coroner reports, and hospital medical records. Exclusion criteria included motor vehicle non-occupants (i.e. pedestrians or motorcyclists) or those with incomplete demographic information. The base-population consisted of persons involved in motor vehicle collisions. The source population consisted of fatal occupants of motor vehicle collisions, and this population was limited through exclusion as defined above, resulting in 80,853 included patients. Of these patients, 67% were not belted and 33% were belted.
Descriptive analyses were performed to assess general trends over the study period among each demographic characteristic. Forward stepwise logistic regression analyses were then performed to estimate adjusted associations between measured demographic characteristics and seat belt use. Odds ratios and 95% confidence intervals (95% CIs) were reported.
Table 1 describes the relative distributions of age, sex, race, safety belt use, and position within the vehicle. Age was categorized as < 25 years, 25 ?60 years, and > 60 years, and race was categorized as Caucasian, African-American, or Other. The majority of subjects were Caucasian male drivers between 25 and 60 years of age who did not use safety belts.
Table 2 describes the results of the logistic regression analyses. Age, sex, race, and position within the vehicle were all determined to be significant predictors for safety belt use as all 95% CIs excluded the null value. Specifically, an increased age, being female, being Caucasian, and being the driver of the vehicle were associated with increased safety belt use in unadjusted and adjusted analyses.
This study has several limitations. Selection bias may have occurred as a result of excluding patients with incomplete demographic data (the authors do not define what proportion of the study population was excluded) or by using only fatally-injured subjects. The use of only fatally-injured subjects limits the generalizability of the study results as passengers not fatally-injured in motor vehicle collisions may have differed significantly from the fatally-injured patients with respect to demographic characteristics or safety belt use. This is a significant drawback of this study although dictated by the design and use of the FARS database. A more valid, but difficult, method may have been to use all subjects of motor vehicle collisions. Using only motor vehicle fatalities to generalize the findings to those who regularly ride in motor vehicles is not very persuasive since fatal subjects may be different, and possibly very different, from those who sustain minor or no injuries. Unfortunately, this is impractical as there are no surveillance systems to identify and collect data on all motor vehicle collisions.
Information bias, in the form of misclassification bias, may have also occurred in this study, although probably to a lesser extent than with selection bias. Misclassification of safety belt use or race, especially due to linkage of demographic data to the FARS database, may have occurred. No attempt was made to identify any form of bias and the authors did not discuss these as potential limitations (another limitation of this study).
Potential confounders, including alcohol or drug use, geographic location, or motor vehicle speed, were not controlled for during the analyses and therefore limited the interpretation of the results. The statistical procedure of forward stepwise regression is problematic in that it is generally thought of as an atheoretical modeling approach that, and depending on the manner in which variables are introduced into the model, may introduce confounding instead of controlling for it. The authors found age, sex, race, and position within the vehicle to be significantly associated with safety belt use using this procedure.
1. National Highway Traffic Safety Administration. FARS database

Feedbacks & Suggestions