Methodological considerations for documenting the energy demand of dance activity: a review

Previous research has explored the intensity of dance class, rehearsal, and performance and attempted to document the body's physiological adaptation to these activities. Dance activity is frequently described as: complex, diverse, non-steady state, intermittent, of moderate to high intensity, and with notable differences between training and performance intensities and durations. Many limitations are noted in the methodologies of previous studies creating barriers to consensual conclusion. The present study therefore aims to examine the previous body of literature and in doing so, seeks to highlight important methodological considerations for future research in this area to strengthen our knowledge base. Four recommendations are made for future research. Firstly, research should continue to be dance genre specific, with detailed accounts of technical and stylistic elements of the movement vocabulary examined given wherever possible. Secondly, a greater breadth of performance repertoire, within and between genres, needs to be closely examined. Thirdly, a greater focus on threshold measurements is recommended due to the documented complex interplay between aerobic and anaerobic energy systems. Lastly, it is important for research to begin to combine temporal data relating to work and rest periods with real-time measurement of metabolic data in work and rest, in order to be able to quantify demand more accurately.


Introduction
The importance of physical fitness in dance and the level of physical fitness required of dancers is a topic of much contention within both dance teaching and training settings and in dance medicine and science literature. While physiological capacity is an important aspect of dance performance, it must also be acknowledged that dance is first and foremost an art form encompassing technical and expressive aspects. Therefore, solely examining the body's physiological adaptations to training cannot infer optimal performance. As stated by Koutedakis and Jamurtas (2004), "dance performance... is a rather complex phenomenon made up of many elements that have direct and indirect effect on outcome" (p. 651-652). Therefore, it could be argued that the presence of an underlying physical fitness foundation is an important pre-requisite to successful and sustained dance performance. Previous literature has highlighted the relationship between fatigue and injury risk with regard to the considerable, and ever increasing, physical demand placed on dancers from choreography in support of this (Koutedakis and Jamurtas, 2004;Allen and Wyon, 2008), further emphasizing the role of appropriate physical preparation in dancers. In sports, training methodologies are based upon in-depth research, which seeks to understand the response of systems of energy utilization during activities and their links to performance capabilities. While dance shares several characteristics with sport there are also fundamental differences, which could make the application of non-dance specific research to dance contexts difficult. With the growth of dance science as a relatively new field of research and its increased dissemination into studio practice, there is a need for evidence to ensure the appropriate preparation of dancers within their training.
The concept of energy demand can be simplified as the energy, or oxygen, cost of completing an activity. Previous research has aimed to classify the energy demand of aspects of dance training (class and rehearsal) and performance, leading to the common description of dance activity as: complex, diverse, nonsteady state, intermittent, of moderate to high intensity, and with notable differences between training and performance intensities and durations (Cohen et al., 1982a;Schantz and Astrand, 1984;Redding and Wyon, 2003;Wyon et al., 2004Wyon et al., , 2011Wyon, 2005;Twitchett et al., 2009a). The diversity of dance performance is particularly highlighted in relation to differences in movement vocabulary and execution both between and within dance genres. One study has noted this diversity as comparable to that of field sports such as soccer and rugby, for which extensive analyses of demand have been undertaken in the past (Wyon et al., 2011). The physiological response of the body to dance activity is noted as complicated and difficult to describe due to these characterizing features, along with a focus on technique and skill.
Only two systematic literature reviews are available in this topic area; one examining fitness components in relation to contemporary dance (Angioi et al., 2009), and one examining existing dance medicine and science literature in Dance Sport (McCabe et al., 2013). Through both of these reviews it is clear that (a) the weak methodologies of many studies do not provide a strong evidence base, (b) studies do not differentiate between levels and styles of dance adequately, and (c) inappropriate comparisons are often drawn between dance and other forms of activity; for example a generalized concept of sport. Other dance specific review articles available include an early paper on physiological aspects of dance (Kirkendall and Calabrese, 1983), a brief review examining classical ballet (Twitchett et al., 2009a), and a review of eccentric muscular contraction in dance activity (Paschalis et al., 2012). Further comment and methodological analysis articles include: examining the dancer as an athlete (Koutedakis and Jamurtas, 2004;Allen and Wyon, 2008), reviewing aerobic and anaerobic aspects of dance (Cohen, 1984), considerations for dance fitness training (Wyon, 2005;Rafferty, 2010), and a methodological review (Redding and Wyon, 2003). While these articles are useful in framing our current understanding of the physiological demands of dance; comprehensive and consensus based conclusions currently do not seem possible largely due to the lack of consistency in methodological design, protocol, and measurement instrumentation, making comparisons between different study findings difficult.
While the above mentioned articles have examined various physiological data in dance populations, to date a comprehensive literature review has not been undertaken examining data available on the intensity and energy demand of various genres of dance. The present study aims to examine the previous body of literature and in doing so, seeks to highlight important methodological considerations for future research in this area to strengthen our knowledge base.

Methods
Pub Med/Medline, Cochrane, and SportDiscus research databases were searched between April and August 2013 with a limit to English language publications utilizing human participants only. Articles published from 1982 to August 2013 regarding the physiological demands and adaptations to dance training and performance were considered. The Medical Subject Heading (MeSH) terms "energy demand, " "physiological intensity, " "aerobic, " "anaerobic, " "movement economy, " "efficiency" were employed in combination with "dance, " "dancer, " "ballet dance, " "contemporary dance, " "modern dance." Articles retrieved were assessed for relevance based on title, abstract, and full-text against the pre-determined inclusion and exclusion criteria. Inclusion criteria were: (I) examining energy demand of dance or the effects of training or performance on the cardiorespiratory fitness of dancers, (II) involving participants defined as professional and/or student dancers in pre-vocational, vocational and/or University settings, (III) examining any professionally recognized genre of dance (for the purpose of this review categorized as ballet, contemporary (incl. modern), or other), (IV) investigating any aspect of dance activity (single exercise, class, training, rehearsal, performance, or competition). Exclusion criteria were: (I) editorials, review or comment articles, and conference proceedings, (II) studies measuring physical fitness variables in a one off screen, (III) studies incorporating aerobic dance (including exercise classes), (IV) studies using dance as an intervention for general population health (i.e., elderly, youth, sedentary, etc.). Finally, reference lists of accepted articles were scanned for further appropriate publications, which were subsequently sourced and included in analysis.
An assessment of study quality of full-text articles deemed relevant was undertaken according to the grading system set out by the ADA evidence analysis library. The grading system, through pre-set criteria, assesses: quality, consistency, quantity, clinical impact, and generalisability. Allowing assigning of grade: I (good), II (fair), III (limited), IV (expert opinion only), or V (grade not assignable). For the purpose of this review, articles graded as expert opinion only, or grade not assignable were omitted from analysis. A flow diagram of the studies identified and included in the review appears in Figure 1. A sample review was completed by an independent reviewer to demonstrate reliability of classification.
Data analysis was conducted on accepted articles to extract relevant information for comparison and construction of evidence tables. These evidence tables form the basis of this review, allowing the summary of previous research findings and highlighting areas in need of further study.

Results
Based on the adopted methodology and search parameters, 30 articles sourced were included in this review; 24 were descriptive studies and six involved an intervention, of which only one was clearly defined as a randomized control trial. Of the 30 articles selected, 10 articles examined ballet; seven contemporary or modern dance; nine were categorized as "other, " which included highland dance, dance sport, folk dance, tap dance, and jazz dance; and four examined multiple genres within the same study, which included ballet, contemporary, and jazz, and additionally included character dance. Terminology used to describe the genre of dance examined varied in some cases. For the purpose of this review the term "contemporary" also includes studies examining modern dance and the term "dance sport-ballroom" also includes studies examining standard or modern ballroom dance. Three studies took their sample from groups of pre-vocational, adolescent dancers, 15 examined professional dancers, 10 examined student dancers, and two looked at both student and professional dancers. Although no universal definition for categorizing an individual as a professional dancer was provided in studies examined, for the purpose of this review the term professional dancer is adopted where participants were described as such within the source study.
Quality assessment undertaken revealed the quality of evidence available to be relatively poor, with 47% graded as "limited" and the remaining 53% as "fair." For the majority of cases, grading within the category "quantity" was the main limiting factor, with a limited number of studies confirming single study findings and low or inadequate sample sizes within studies.
Results are organized by aspects of dance activity measured in order to accurately represent different elements of a dancer's schedule. Five studies undertook measurements on more than one aspect of dance activity and therefore are included under multiple sub-headings (and appear twice or more in subsequent counts).

Energy Demand of Class
Literature examining dance class generally classifies the energy demand as moderate to high and intermittent in nature; although a greater intensity and shorter duration of exercise is noted in the center phase compared to the warm-up.
Thirteen papers in total have investigated elements of dance class in various genres. Four measured only the execution of a single exercise within a class setting ( Table 2) (Guidetti et al., 2007a(Guidetti et al., , 2008Oliveira et al., 2010;Maciejczyk and Feć, 2013), with the remaining nine examining an entire class, often drawing comparisons between different sections; mostcommonly the warm-up and center phases ( Table 1) (Cohen et al., 1982a;Schantz and Astrand, 1984;Rimmer et al., 1994;Dahlstrom et al., 1996;Dahlstrom, 1997;Wyon et al., 2002Wyon et al., , 2004Baillie et al., 2007;Guidetti et al., 2007a). Of the 13 papers examined, the majority of studies examined either ballet or contemporary dance with single, small scale studies additionally examining highland, tap, and folk dance. Groups of prevocational adolescents, undergraduate and graduate students, and professional dancers were examined. Data was reported on a variety of variables as displayed in Tables 1, 2. More detailed information relating to specific methodologies of each study are available in Supplementary Table 1. Reported work and rest temporal data suggests dance class is an intermittent form activity. Differences are noted between the work to rest ratio of the warm-up/barre phase of class and that of the center/execution phase, with the center/execution phase consisting of shorter work and longer rest periods (Schantz and Astrand, 1984;Dahlstrom et al., 1996;Wyon et al., 2002Wyon et al., , 2004. Though methodologies and variables of focus vary, the magnitude of physiological strain is described throughout the breadth of literature as moderate to high, although this is also seen to differ between different sections of the class. Mean intensity (as represented by reported aspects of VO 2 , HR and EE) of the warm-up/barre phase is consistently reported as lower than that during the center/execution phase (Cohen et al., 1982a;Schantz and Astrand, 1984;Dahlstrom et al., 1996;Dahlstrom,   1997; Wyon et al., 2002Wyon et al., , 2004. This difference was reported as significant for both VO 2 and HR data in four of the studies presented (Cohen et al., 1982a;Schantz and Astrand, 1984;Wyon et al., 2002Wyon et al., , 2004. Three studies have compared the intensity of single exercises within a class to individual lactate/anaerobic/metabolic acidosis threshold, and described demand as working close to (Maciejczyk and Feć, 2013), often above (Guidetti et al., 2007b), 10% less than and non-significantly different to (Oliveira et al., 2010) threshold levels in Folk, Ballet, and Tap dance, respectively. Largely negligible differences in the intensity of class can be seen between dance genres, when visually comparing different studies. Only one study reported direct comparisons between classes of different dance genres and found no significant differences in the median HR-values (Dahlstrom et al., 1996). One study reported no significant differences between classes in different techniques of the same genre . Lastly differences between participant characteristics such a gender and training status are reported. For example, significant differences have been reported between sexes in O 2 requirement for the whole class, warm-up, and center phases . One study noted significant differences between percentage of dancetime (work-time) and mean whole class HR-values between undergraduate, graduate and professional dancers (Wyon et al., 2002). Another study found no significant difference in overall energy requirement of a single ballet exercise between dancers ranked as high or low technical ability, however did find significant differences in energy source contribution with high technical ability dancers using significantly higher aerobic source contribution and significantly lower anaerobic lactic system contribution than the low technical ability dancers (Guidetti et al., 2008).
The data drawn from previous research examined here provides basic information regarding the intensity of dance class and therefore alludes to the demand placed upon dancers during class. However, it should be noted that data are rarely presented relative to the individual dancer's fitness levels.
Significant differences are noted between different phases of class, the response of different sexes and of dancers of varying technical ability, which must be taken into account during the interpretation of data sets.

Energy Demand of Rehearsal
Very few studies have presented data on the energy demands of rehearsal and high variation is reported between results of those that have. Therefore, an ambiguity regarding the demands of rehearsal prevails.
Measurements during rehearsals have been undertaken in ballet (Schantz and Astrand, 1984;Rimmer et al., 1994), contemporary , and highland dance (Baillie et al., 2007), with samples taken from student and/or professional dancers. Two of the papers included in this section of the review claim to also measure performance data, however Schantz and Astrand (1984) do not differentiate between measurements that took place in final rehearsals and those conducted during performance and Wyon et al. (2004) used measurements undertaken within dress rehearsals as their performance data. Therefore, all data described above is included here as measuring the physiological demands of dance rehearsal. Data available from these studies is reported in Table 3, with more detailed information relating to specific methodologies available in Supplementary Table 2.
Reported mean HR-values for female participants range from 108 to 176.9 b.min −1 across two studies Baillie et al., 2007), with the third reporting HR-values "frequently close to max" (Schantz and Astrand, 1984), and the final study reporting that dancers spend 52% of the rehearsal time working at 60-90% HR max (Rimmer et al., 1994). One study described the structure of rehearsals stating that segments of dance were performed at a high intensity level followed by a period of rest (Rimmer et al., 1994). One study reported significant differences for both mean O 2 requirement and mean HR between rehearsal and dress rehearsal measurements, with no significant differences between sexes .

Energy Demand of Performance
Through literature examined the intensity of dance performance is almost unanimously described as high/heavy, with an intermittent nature, utilizing both aerobic and anaerobic energy systems. Eleven papers have measured aspects of performance or competition, although the conditions under which measurements were taken do vary ( Table 3). Five studies involved simulated competition, mimicking temporal characteristics of typical competition format, in Dance Sport (Latin American or Ballroom) (Blanksby and Reidy, 1988;Kļonova and Kļonovs, 2010;Bria et al., 2011;Massidda et al., 2011;Liiv et al., 2013). Three studies have undertaken real-time measures; one during a scheduled championship competition (Baillie et al., 2007), one during stage performance (Cohen et al., 1982b), and one including readings immediately postperformance (Galanti et al., 1993). Two studies undertook retrospective analysis from video recordings of performances (Twitchett et al., 2009b;Wyon et al., 2011), and the remaining study does not specifically outline the conditions of the performance measures (Redding et al., 2009). The majority of these studies examined professional dancers, with only one study taking its sample from student dancers. Data were reported on a variety of variables as outlined in Table 4. More detailed information relating to specific methodologies of each study are available in Supplementary Table 3.
The data available suggests a high mean intensity of dance performance/competition across all genres, with participants frequently reported as reaching over 80% of their HR max (Cohen et al., 1982b;Blanksby and Reidy, 1988;Galanti et al., 1993)and/or VO 2max (Blanksby and Reidy, 1988;Liiv et al., 2013). However, when looking more in-depth at the data available, there seems to be no agreement regarding the amount of variation both within and between performance repertoires examined. One study found significant variations in percentage time spent working at different intensities among different positions within a classical ballet company, with soloists spending a higher percentage of total time resting than principals and principals spending a higher percentage time working at moderate intensity than soloists and artists (Twitchett et al., 2009b). The same study commented on little variation between demands of different classical repertoire examined (Twitchett et al., 2009b). Similarly, when comparing end-performance blood lactate levels in contemporary dancers, no significant difference was noted in values from four different pieces of repertoire, with all measurements noted as under 4 mmol.L −1 (Redding et al., 2009). Conversely, one study found significant differences in blood lactate levels between the three dance techniques performed in highland dance (p < 0.01) (Baillie et al., 2007). In a study comparing performance repertoire of two dance genres (ballet and contemporary/modern), significant differences were found between genres for time (s.min −1 ) spent at all subjectively categorized exercise intensities (excluding "hard"). For example, significantly more time was spent at "rest" and "very hard intensities" during ballet than contemporary performance (p < 0.001) (Wyon et al., 2011). Based on data presented, it is currently unclear whether any generality can be drawn in the intensity of performance between or within dance genres.

Impact of Training/Performance on Cardiorespiratory Fitness
Data presented suggests that a program of dance activity alone is insufficient to elicit improvement in cardiorespiratory fitness, with the exception perhaps of a prolonged period of performance. This is suggested as causal of the low aerobic fitness levels often reported in dancers.
Nine studies examined adaptation of elements of cardiorespiratory fitness longitudinally over a range of training approaches. Four studies did not alter the training undertaken and simply measured at multiple time-points to see the effect of a typical training/performance period (Dahlstrom et al., 1996;Koutedakis et al., 1999;Wyon and Redding, 2005;Martyn-Stevens et al., 2012) (Table 5). However, five studies additionally involved intervention in the dancers' usual training schedule, by introducing fitness training programmes/sessions, or extra classes/rehearsals into the dancer's schedules (Galanti et al., 1993;Ramel et al., 1997;Koutedakis et al., 2007;Angioi et al., 2012;Mistiaen et al., 2012) (Table 6). More detailed information relating to specific methodologies of each study are available in Supplementary Table 4.
Of the four studies that did not implement an intervention to the schedules and/or training of the dancers differences can be observed in available data related to the characteristics of the dancers schedules at the time of testing, for instance between periods of rehearsal and periods of performance (Wyon and Redding, 2005). Over the course of a 3 year dance training program, with aerobic fitness measured four times per year, a significant change in predicted VO 2max was only observed between the first and second half of the third year of study (Dahlstrom et al., 1996). A sub-set of participants who completed measurements at all-time points recorded an mean 20% increase over the 3 years (Dahlstrom et al., 1996). Two studies examined changes in cardiorespiratory fitness during a narrower time frame of a specific rehearsal and performance cycle/season, with contrasting findings (Wyon and Redding, 2005;Martyn-Stevens et al., 2012). One study undertook a VO 2max -test and the Wingate anaerobic bike test 10-12 weeks prior to a single performance and then again one-to-two weeks following the performance, reporting significant increases in absolute and relative peak power output and fatigue index, but no significant changes in VO 2max -values (Martyn-Stevens et al., 2012). Whereas, significant decreases in HR-, %HR max -, and BLavalues at stage 5 of the DAFT were observed following an 8 week performance period (tour), indicating enhanced aerobic fitness (Wyon and Redding, 2005). In the same study, no significant differences were reported during a 12 week rehearsal period leading up to the performance period (Wyon and Redding, 2005). The fourth study measured changes in VO 2max and anaerobic power over the course of a 6 week rest/holiday period and over the course of the subsequent two-to-three month training period. During the rest period significant increases were reported in peak anaerobic power (p < 0.01) and VO 2max (p < 0.05), with   (Koutedakis et al., 1999). Across all nine studies, mean VO 2max -values reported ranged from 37.4 (± 4.1) to 56.6 (± 9.3) ml.kg.min −1 , with the top end of the range representing data recorded immediately postaerobic training intervention. Furthermore, all five studies that implemented an intervention into the typical training schedules of the dancers, reported significant increases in measures of aerobic fitness in participants exposed to the intervention (training/exercise groups), despite differences in intervention design and length, and exercise protocol employed. One of these studies included only additional dance rehearsals, with no supplementary fitness intervention and noted improvements of 15% in a sub-maximal measure of VO 2peak (Galanti et al., 1993).
Of the three studies that additionally undertook measurements with a control group, two also reported no significant changes in the mean VO 2max of control subjects (Ramel et al., 1997;Koutedakis et al., 2007), with one reporting no increase in control subjects' mean HR at DAFT stage 5 (Angioi et al., 2012). Finally, of note are the corresponding significant increases in measured aspects of dance ability/competence found only in subjects exposed to training intervention in two of the studies (Koutedakis et al., 2007;Angioi et al., 2012).

Discussion
Evidence presented in the above review pertains to either the physiological demand of dance activity (within class, rehearsal  or performance) or the longitudinal tracking of measures of aerobic and/or anaerobic fitness. Significant differences are noted in results of multiple variables presented, which are further discussed herein.

Physiological Demand of Dance Activity
As demonstrated by results of studies included in this review, the classification of the degree of physiological strain placed on the body during dance class involves a complex interaction of the intermittent work and rest periods and the varying intensities of work depending on specific movement vocabulary executed at different stages of the class. With most studies noting differences between two distinct phases that exist within dance classes of multiple genres, it is suggested that mean data based on the entire length of a class should be discarded. Wyon et al. (2002) recommended that classification as a high intensity intermittent activity should only be applied to the center phase work of the dance class. Furthermore, this study highlighted important influences on the total time spent dancing, such as the amount of time spent by the teachers correcting technique and the size of each group (Wyon et al., 2002). It is evident from the review of literature that very few studies have examined dance rehearsal and amongst the data available variation is too high to allow a generalized statement of the energetic demand. In a review, Wyon (2005) noted the diversity in physical stresses previously reported and suggested the presence of a build in intensity of rehearsal closer to performance. As previously noted, a study by Wyon et al. (2004) reported significant differences for both mean O 2 requirement and mean HR between rehearsal and dress rehearsal measurements. While the length of time between the two measurements is not stated in this instance, this nevertheless supports the suggestion that there is likely to be an influence of rehearsal status on values. It is recommended that future studies give a detailed account of the time frame in which measurements were taken (i.e., how far along the rehearsal process) and describe the temporal characteristics of the rehearsal, including time spent learning, marking, correcting, and completing full-run through's at maximum effort. Full-run through intensity is likely to closely mirror that of performance, based upon data presented by Wyon et al. (2004). Conversely, Schantz and Astrand (1984), while not differentiating full data sets, noted heart rate values 5-10 b.min −1 lower and blood lactate concentrations 8% lower during measurements without audience in rehearsal compared to those recorded during performance of the same dance piece in the same dancers.
While the high intensity nature of performance across dance genres is evident from data sets presented, there remains disagreement as to the diversity of different dance repertoire within single dance genres. On the one hand, there is the suggestion that performance demands vary considerably (Wyon, 2005;Baillie et al., 2007) and on the other hand, there is the suggestion that physiological demands rarely change dramatically with different repertoire (Redding et al., 2009;Twitchett et al., 2009b). Only one study has carried out direct comparison between a range of performance repertoire of different dance genres, namely ballet and contemporary, and, as previously stated, reported significant differences in time spent working at a range of exercise intensities (Wyon et al., 2011).
The main problem faced in dance training, as highlighted by the data presented, is the significant differences apparent between the energetic demand of class, rehearsal, and performance. This, along with the lack of documented positive physiological adaptation occurring through periods of dance training and rehearsal, has often lead to the conclusion that dancers are not physically prepared for performance Wyon and Redding, 2005). It is also important to acknowledge the relative nature of intensity measurements, such as those undertaken by literature presented. Wyon (2005) highlighted that while the absolute intensity of a movement sequence, be it executed as part of class, rehearsal, or performance, is set by the teacher/choreographer, the relative intensity recorded is largely influenced by the fitness of the individual dancer. It stands to reason that an individual with greater cardiorespiratory fitness would have an increased capacity to cope with the demands being placed on them during movement, therefore resulting in a documented lower relative intensity. This could be particularly important to consider in the analysis of performance data in light of suggestions that dancers do not possess the adequate fitness to cope with performance demand. Furthermore, Allen and Wyon (2008) suggest that dancers of a high skill level are likely to possess a very good economy of movement, reducing the physiological stress dancing places on them.

Impact of Training/Performance on Cardiorespiratory Fitness
While studies have commented on the possible cardiorespiratory training effect of dance class based on total length of time spent in an appropriate training zone (suggested as 60-70% HR max ) throughout the class, later studies have criticized this conclusion due to the small continuous time spans during which these values are attained. For example, Rimmer et al. (1994) noted that HR was not elevated into a training zone for longer than 1-6 min at any given time.
In terms of rehearsals being able to elicit positive physiological adaptation, Redding and Wyon (2003) speculated... "often rehearsal periods are short and do not allow enough time for the physiological adaptation to take place that would enable the dancer to cope with the increased demands of performance" (p. 10). Two studies have tracked change in fitness variables throughout a rehearsal period and noted no significant differences on aerobic fitness measures of VO 2max or HR at stage 5 of the DAFT (Wyon and Redding, 2005;Martyn-Stevens et al., 2012), but a significant difference in measures of anaerobic power (Martyn-Stevens et al., 2012). A complex interplay between time and intensity is again perhaps a limiting factor in aerobic adaptation, with the emphasis of rehearsals remaining on skill acquisition and learning.
The high intensity nature of performance has prompted speculation from research as to the possible training effect of performing, with suggestions of eliciting an aerobic training response (Blanksby and Reidy, 1988;Twitchett et al., 2009a). Intensities and durations during ballet performance have been noted as baring similarities to recommendations for interval training, with the frequency of performance instead suggested as potentially limiting factor to positive adaptation (Cohen et al., 1982b). No studies to date have clearly documented performing frequency, although one study by Wyon and Redding (2005) stated an 8 week performance period for two separate contemporary dance companies and a significant change in stage 5 DAFT HR's over this time.
Control groups of intervention studies have typically involved a sample of dancers who continue their typical training schedule, or in other words are subjected to a "dance only" program. Over a 12-week period Koutedakis et al. (2007) noted no increase in VO 2max under this training condition, compared to a significant increase in the sample undertaking additional supplementary fitness training. A further study included an additional contemporary dance technique class for those in the "dance only" condition and reported a non-significant change in markers of aerobic fitness (Angioi et al., 2012). Moreover, all studies included in the present review that involved supplementary fitness training intervention noted positive cardiorespiratory adaptation in exercise groups regardless of design, length, and protocol (Ramel et al., 1997;Koutedakis et al., 2007;Angioi et al., 2012;Mistiaen et al., 2012). From the literature currently available, it can subsequently be concluded that in order to elicit positive adaptation in markers of cardiorespiratory fitness, supplementary fitness training is required. Wyon and Redding (2005) further suggest that: "Cardiorespiratory training needs to be planned and managed to the same extent as the rehearsal schedule to allow the dancers to peak for each performance period as physically as they do technically" (p. 74).

Methodological Limitations
Methodological limitations of the examined literature fall under four main categories: use of mean data, prediction of VO 2 from previous steady-state work, validity and reliability of fitness test protocols, and variables of focus.
Firstly, with all modes of dance activity described as intermittent and non-steady state in nature, the reporting of mean data is deemed somewhat invalid. As early as 1983, a review highlighted the potential underestimating of caloric expenditure in dance due to the inclusion of rest periods reducing the energy expenditure when calculated per unit of time (Kirkendall and Calabrese, 1983). Redding and Wyon (2003) also stated that "dance research that has attempted to measure work output in class, rehearsal, and performance often appears not to have considered two factors: first, that dance is not a steady-state activity... and second, that mean calculation of work output probably does not provide particularly useful information" (p. 15). However, studies have continued to report data in this way, commenting on the potential influence on results in some cases but not removing the limitation from their analyses. It is of interest to document the intensity of work periods and that of rest periods separately, as well as examine response to transitory periods, to allow an understanding not only of work intensity, but also of features of recovery as important markers of efficiency and cardiorespiratory fitness.
Secondly, the use of HR-VO 2 relationships established during previous steady-state treadmill work to enable prediction of VO 2 from HR's measured during dance activity, as adopted by some previous research, has been invalidated by data supplied by two studies. Schantz and Astrand (1984) found the prediction of VO 2 in ballet class to be under-estimated by 3% (range: -9 to 21) during barre exercise, by 7% (range: -9 to 21) during moderate center work, and by 15% (−1 to 21) during severe center work. The findings of a later paper by Redding et al. (2004) further confirmed these results, reporting differences of up to 30 b.min −1 between the treadmill and dance protocols at any given VO 2 -value and an increase in variability with increasing intensity, up to differences of 49 b.min −1 at a VO 2 of 60 ml.kg.min −1 .
Thirdly, the validity, reliability, and sensitivity of adopted protocols for assessing cardiorespiratory fitness are questioned. For instance, the importance of weight bearing and muscle group activation has been highlighted (Redding and Wyon, 2003), suggesting that tests undertaken on a cycle ergometer, for example, may be less valid. The validity of prediction of maximal fitness from sub-maximal testing protocols is also questioned. Redding and Wyon (2003), in a review of aerobic testing methods, highlighted a necessary trade-off between reliability and validity in field predictions with potentially greater error vs. lab based controlled tests. There may also be an influence of the timing of testing, as demonstrated by significant differences in VO 2max , before and after a 6 weeks holiday and again after a preparatory training period (Koutedakis et al., 1999). Martyn-Stevens et al. (2012) also highlighted a possible influence of familiarity with the test protocols in improvements recorded during post-testing.
The choice of variables reported largely influence the conclusions of the studies presented. Cohen et al. (1982a) highlighted the challenge of measuring absolute intensity of dance activity, due to an inability to accurately measure work output, as well as difficulties in relative measurements due to variability in the level of the dancer (time spent training/skill level) and the type and style of dance. Potential influences of muscular contraction type on metabolic data outputs have also been cited. Guidetti et al. (2007a) speculated that during a ballet class, heart rate may be more closely associated with skeletal muscle contraction than the level aerobic demand, with a potential influence of the combination of static and dynamic leg and arm movements. Paschalis et al. (2012) further cited potential influences of eccentric muscular contraction during dance movement on resting energy expenditure and movement economy. Lastly an overreliance on reporting VO 2max as a marker of cardiorespiratory fitness is noted. Guidetti et al. (2007a) commented on the importance of relating measurements to individual dancers ventilatory and anaerobic thresholds as these proved more sensitive to detect differences between groups of varying skill levels compared to %VO 2max or %HR reserve or absolute HR. Given the high intensity nature of dance, measures relating to ventilatory and/or anaerobic and/or lactate thresholds seem valid, however have been examined by very few studies to date.
In addition, it is important to account for the conditions under which data claiming to document the intensity of performance or competition is collected. One study utilizing dress rehearsal data to represent performance was previously highlighted as potentially invalid and therefore analyzed within this review as rehearsal data . The use of simulated competition rather than actual competitive conditions in the available Dance Sport research (Blanksby and Reidy, 1988;Kļonova and Kļonovs, 2010;Bria et al., 2011;Massidda et al., 2011;Liiv et al., 2013) also perhaps warrants the same scrutiny. It is important to note the restriction placed on performance measurements by equipment, whereby only equipment that can be hidden under costumes is typically utilized in live performance or competition. This has lead to an over-reliance on HR data; the questionable validity and reliability of which has already been discussed. It may therefore be pertinent for future research to examine the ability of carefully designed simulated conditions such as these to represent performance and/or competition demands with an adequate degree of accuracy.

Conclusions and Future Recommendations
Most individual studies examined within this review claim that dance can be classified as high intensity, intermittent exercise. Based on the temporal data examined within this review, it can be agreed that dance is an intermittent activity; however data is highlighted as inconsistent in terms of intensity, with large variability reported between study results. Furthermore, without a clearer conclusion regarding the intensity of dance training and performance, the levels of cardiorespiratory fitness required by dancers remain relatively unknown, although it has been previously suggested that "contemporary dancers need to be both aerobically and anaerobically fit in order to be prepared for the many different demands of the genre" (Redding et al., 2009, p. 8).
Previous reviews have commented on barriers to a consensual conclusion and suggest further research is necessary. Specific recommendations to guide future research, based upon the findings of this review, are as follows. Firstly, research should continue to be genre specific, with detailed accounts of technical and stylistic elements of the movement vocabulary being examined. Secondly, a greater breadth of performance repertoire, within and between genres, needs to be closely examined as a fuller understanding of demand, as well as similarities and/or differences of measurements, is necessary to determine appropriate cardiorespiratory fitness levels for dancers. Thirdly, a greater focus on threshold measurements is recommended due to the documented complex interplay between aerobic and anaerobic energy systems. Lastly, it is important for research to begin to combine temporal data relating to work and rest periods with real-time measurement of metabolic data in work and rest, in order to be able to quantify demand more accurately.

Supplementary Material
The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fpsyg. 2015.00568/abstract