Classical Ballet for Women Aged Over 50 Years: Investigating Balance, Strength, and Range of Motion

ABSTRACT

Background: Regular exercise can mitigate the loss of strength, balance, and flexibility that contribute to age-related decline in physical function and mobility. However, traditional exercise interventions often report poor adherence rates. There is growing interest in classical ballet as an enjoyable exercise modality for adults in middle to late age. Classical ballet requires muscular strength, coordination, and flexibility. The current study investigated a classical ballet intervention on the balance, physical function, and range of motion of women aged over 50 years. Methods: Twenty-two healthy female participants (aged 56.2 (4.5) years (mean (SD)) completed a 10-week ballet intervention. Results: This single-arm study showed significant improvements (p<.05) in lower limb strength (measured by 5 times sit-to-stand and forward leap) and high adherence rates (95% adherence for participants who completed the intervention). No adverse events were reported. Improvements in balance were reported in the left leg only (as measured by center of pressure ellipse area in the parallel retiré condition). Conclusions: These results allude to the positive effects of ballet training on strength and balance in adults aged 50 years and over. High adherence rates suggest that ballet training was enjoyed and may thus be a long-term exercise modality for this population. Although this study was a single-arm design, it suggests promising results for future research wishing to evaluate the effectiveness of classical ballet training using randomized controlled trial designs.

KEYWORDS:

• Aging
• balance
• physical activity
• strength

Classical ballet is a dance style and art form based on traditional esthetic requirements and principles of technique, which focus on specific positions of the upper and lower limb. High levels of balance, strength, and flexibility are required to complete fundamental movements of this technically demanding style of dance (Kadel et al., 2005; Khan et al., 1997; McCormack et al., 2019). To demonstrate an elite level of classical ballet, many years of dedicated training, typically from a young age, are required. However, there is a growing interest in recreational ballet classes for older adults, where social inclusion, self-expression, and increased physical activity levels are seen as some of the primary drivers for participation (Fong Yan et al., 2018; Houston & McGill, 2013; The Royal Academy of Dance, 2017a).

The aging process leads to the deterioration of physiological systems, which results in deleterious effects on functional ability, including reduced range of motion (ROM) and strength, and poor balance (Chodzko-Zajko et al., 2009). The current evidence-based literature unequivocally recommends participation in structured exercise for older adults to minimize the progression and impact of age-related physiological and functional deteriorations (Chodzko-Zajko et al., 2009; Garber et al., 2011; Mattle et al., 2020; Taylor, 2014; Tricco et al., 2017). Engaging in regular exercise training in middle age is also associated with healthy aging, thus habitual training in middle—older aged adults may assist to preserve physical function and slow the rate of decline in older years (Sabia et al., 2012; Sun et al., 2010). However, traditional exercise interventions often report poor adherence rates (Room et al., 2017; Skelton & Todd, 2004) which have been attributed to the view that traditional exercise training is not enjoyable (Chao et al., 2000). To gain maximum benefit and maintain adaptation from training, exercise must be performed regularly (Tokmakidis et al., 2009). If an exercise program is perceived to be unenjoyable, long-term adherence is unlikely (Collado-Mateo et al., 2021; Linke et al., 2011; Ryan & Deci, 2000). Thus, alternatives to traditional exercise programs are needed to maintain long-term adherence.

Previous research has demonstrated the efficacy of dance training to improve balance, muscular strength, power, and aerobic capacity (Hwang & Braun, 2015; Keogh et al., 2009; Rodrigues-Krause et al., 2019). Individuals who participate in dance interventions report high levels of enjoyment and exhibit adherence rates which are the same as, if not better than, that of traditional exercise (combined aerobic and resistance) programs (Fong Yan et al., 2018; Hwang & Braun, 2015).

Regular participation in classical ballet includes training in balance, flexibility, whole-body ROM, and coordination (Lin et al., 2014) which can require high physiological demands. Therefore, this type of training may elicit positive physiological adaptations and assist to maintain physical functioning for middle—older aged adults. Due to the limited existing research, promising findings have emerged from novices participating in classical ballet training. Mixed-genre interventions incorporating classical ballet have demonstrated improved perceptions of quality of life and increased physical activity levels to meet recommended guidelines (The Royal Academy of Dance, 2017b) as well as greater balance and high motivation to attend classes with a 100% adherence rate (Houston & McGill, 2013). Classical ballet interventions designed for older adults with Parkinson’s Disease have reported positive impacts on physical, social, emotional, and cognitive health following classical ballet training (Queensland Ballet, 2014, 2018). Additionally, The Royal Academy of Dance (RAD) has designed the Silver Swans program, which is targeted at older learners and aims to improve ROM, posture, co-ordination, and energy levels (The Royal Academy of Dance, 2017b).

Given the high adherence rates previously reported in dance-based exercise (Fong Yan et al., 2018) and the potential for classical ballet training to elicit positive physiological adaptations (Queensland Ballet, 2014, 2018), ballet training may be a sustainable modality for adults to remain physically active long term. However, to date, there is limited published peer-reviewed literature on the effects of classical ballet training for middle—older aged adults (Letton et al., 2020). This study aimed to evaluate a classical ballet training intervention for adults aged 50 years and over on improving balance, physical function, and ROM.

Material & methods

Study design and sample size calculation

A single-arm pilot study design was used. An a priori sample size calculation was obtained using G*Power version 3.1.9.4 (Faul et al., 2007). Based on the primary outcome measure, balance, using previously obtained data (Shigematsu et al., 2002), a sample size of 21 participants was required for an effect size of 0.75 and a power of 0.95. To allow for attrition, the current study aimed to recruit 26 participants.

Recruitment and inclusion criteria

The study was approved by the University Human Ethics Committee. Participants were recruited through e-mail advertisements distributed to university staff and affiliates. All participants provided written informed consent before completing a medical history questionnaire to determine eligibility. Responders aged 50 years and over who were not currently participating in classical ballet classes and not currently receiving treatment for any diagnosed chronic health condition or experiencing musculoskeletal injury or disability which would prevent them from dancing were deemed eligible. Participants with prior ballet training experience were considered eligible. Participants were excluded if they were unable to attend the scheduled weekly ballet class or unable to complete basic ballet movements unaided. Age requirements for inclusion in the current study were guided by the protocol for the Silver Swans program from the RAD (The Royal Academy of Dance, 2017b). In addition, the current study focused on the potential for classical ballet training to prevent age-related physical decline, necessitating recruitment of middle-aged adults of at least 50 years.

Experimental protocol

Participants completed questionnaires, functional, balance, and ROM assessments pre- and post-intervention. All assessments were conducted in the same order by the same investigator at both time points. Where multiple trials were attempted (as specified by previously published assessment protocols), approximately 1-min rest was allowed between attempts, and the best score was used. Participants were given the opportunity for one practice trial. Participant demographic and anthropometric data, including age, sex, and body mass index (BMI) were also recorded.

Questionnaires

Three validated and easy to administer questionnaires were selected: The International Physical Activity Questionnaire (IPAQ) (Hagströmer et al., 2006) Self-Efficacy for Exercise (SEE) scale (Resnick & Jenkins, 2000), and Rosenberg Self-Efficacy (RSE) scale (Rosenberg, 1965). These questionnaires provided self-reported data pertaining to habitual physical activity levels, perceived confidence to exercise in a variety of conditions, and general self-esteem, respectively.

Functional assessments

Physical function was evaluated using aspects of the Senior Fitness Test (Langhammer & Stanghelle, 2015). Specifically, the timed up and go (TUG) (mobility), chair sit and reach test (lower limb and back flexibility), and Apley back scratch test (shoulder flexibility) were used. The protocol for these tests has been previously validated elsewhere (Langhammer & Stanghelle, 2015). The five times sit-to-stand was used to assess lower body strength (Bohannon, 2011). Handgrip strength (kg) was measured during a maximal grip strength test on the dominant hand, using a hydraulic hand dynamometer (Jamar, Model: FE-12-0600, Sammons Preston Rolyan, Bolingbrook, IL, USA). The dominant hand was assessed as its strength is commonly reported to be greater than that of the non-dominant hand (Incel et al., 2002). Quadriceps strength (kg) of the dominant leg was measured during a maximal seated isometric contraction (Desrosiers et al., 1998) using a spring gauge (Art. No 29,942, Albert Kerbl GmbH, Buchbach, Germany). Dynamic balance was assessed using the forward functional reach test (Duncan et al., 1990).

Three-dimensional optical motion analysis

Optical motion capture was used to assess lower limb joint ROM, strength, and static balance during ballet specific movements. Data from an eight-camera (Vicon Vantage) optical motion capture system (100 Hz; Vicon Motion Systems, Oxford Metrics, UK) and two in-ground force plates (1000 Hz; AMTI, Watertown, MA, USA) were collected simultaneously via Vicon Nexus 2.8 software. Participants were fitted with 39 retroreflective markers, attached bilaterally to bony and anatomical landmarks according to the Newington-Helen Hayes and Vicon Plug-in Gait model (Davis et al., 1991; Kadaba et al., 1990; Woltring, 1986). The Dynamic Plug-in Gait model was used to process all trials. Motion analysis data were filtered using the Woltring Quintic Spline Interpolation filter, rigid body gap fill, pattern gap fill, and Woltring Trajectory Filter. Trials that were unable to be adequately reconstructed and gap filled were recaptured.

Static balance was assessed using four single-leg balance (SLB) variations. Each variation was performed on both left and right legs. Participants were instructed to maintain the SLB position for as long as possible and keep arms by their side throughout the test. The trial concluded if the participants achieved a balance duration of 30 s or if they were unable to maintain their position. The four static balance variations performed were:

1. Functional SLB with eyes open: standing upright with legs straight and in a parallel orientation. The knee of the lifted leg was flexed to 90 degrees (Briggs et al., 1989; Franchignoni et al., 1998)

2. Functional SLB with eyes closed: as above, but with eyes closed (Briggs et al., 1989; Franchignoni et al., 1998)

3. Turnout SLB: standing upright with legs turned out and hip externally rotated (the balletic first position). The knee of the lifted leg was flexed so that the foot of the lifted leg touched the knee of the standing leg (the balletic “turned out retiré” position)

4. Parallel SLB: standing upright with both legs in parallel. The knee of the lifted leg was flexed so that the foot touched the knee of the standing leg (the balletic “parallel retiré” position)

Lower limb ROM of the hips, knees, and ankles was assessed using five ballet-specific movements (Figure 1). Hip flexion was assessed using (i) forward kick (grand battement devant from the feet in first position) and hip extension was assessed using (ii) backward kick (grand battement derriere from the feet in first position). Participants were instructed to kick to the end of their comfortable range (i.e., as far as they could go safely) either forward or backward, respectively. Knee flexion was measured using (iii) deep squat (grand plié in second position). Participants were instructed to bend their knees as far as possible while maintaining an upright torso and feet flat on the ground. Finally, ankle dorsiflexion was assessed using a (iv) shallow squat (demi plié in first position) and ankle plantar flexion was assessed using (v) calf raise with the legs in a parallel orientation. Standardized verbal instructions and demonstrations were given. Assessments were modified where necessary for participants based on individual ability.

Figure 1. Movements used to assess joint ROM: (i) forward kick (grand battement devant from feet in first position); (ii) backward kick (grand battement derriere from feet in first position); (iii) deep squat (grand plié with feet in second position); (iv) shallow squat (demi plié with feet in first position), and (v) calf raise with feet parallel.

Lower limb strength was assessed using three movements: (i) maximal vertical countermovement jump (sauté), with feet in parallel and arms relaxed by the side; (ii) vertical countermovement hop; and (iii) forward leap with countermovement. Participants were instructed to jump, hop, or leap as high or as far as safely possible during each trial, respectively.

Data analysis

To analyze three-dimensional optical motion capture data, the start and peak of each movement were identified using objective criteria. For ROM trials, joint angle was calculated as joint angle at maximum depth or peak of the movement minus joint angle at the start of the movement (degrees). For jump and hop trials, the difference in vertical elevation of the pelvis at the start of the movement minus that at the end of the movement was calculated. For leap trials, the difference in anterior displacement of the foot at the start of the movement minus the position of the foot of the contralateral limb upon landing was calculated. The customized code was written in MATLAB (MathWorks; Natick, MA, USA) and run via integration with Vicon Nexus software to automate this process.

Specific criteria were also used to identify the start and end of balance trials. The start of each trial was defined as the time at which the foot of the non-supporting leg was lifted to reach its balancing position, and a plateau in vertical center of mass (CoM) was established. The trial was terminated when balance was lost, indicated by the foot of the lifted leg beginning to descend, which was marked as the start of the decrease in vertical CoM, or when the eyes opened (during eye closed tests only). A maximum time of 30 s was recorded. Balance outcomes recorded were as follows: (i) total duration (seconds), (ii) the 95% confidence ellipse area of the center of pressure (CoP) in both the anterior/posterior (A/P) and medial/lateral (M/L) directions and reported as a percentage of the inter-anterior superior iliac spine (IASIS) distance squared (%IASIS) (Khan et al., 1997); (iii) CoP-CoM distance in both the A/P and M/L directions (%IASIS) (Khan et al., 1997); and (iv) average CoP velocity in both the A/P and M/L directions and reported as percentage of IASIS, per second (%IASIS/s).

Ballet intervention

The intervention consisted of one 60-min ballet class once per week, for 10 weeks. The study design consisted of one class per week as this was considered to be reflective of a realistic and achievable real-life commitment for this population. Two different class time options were offered each week, either Monday or Wednesday, to accommodate participants’ varying schedules. All classes were run by the same instructor and followed the same plan. Participants were required to attend a minimum of eight classes across a period of 13 weeks, to allow for adequate attendance while accounting for misadventure and scheduled absences (e.g., public holidays). Classes were instructed by a registered teacher of the RAD and licensed RAD Silver Swans instructor, who is also a member of the research team (REW). The instructor did not contribute to any direct assessment of outcome measures with participants, but only processing of deidentified balance and ROM digital data. Each class adhered to typical basic ballet class structure, comprising a general 5-min warm up, 50-min ballet classes (including exercises performed at the barre and in the center, described below) and a 5-min cool-down. Classes were progressed in difficulty as participants’ ability progressed. Movements were modified, where necessary, to fit individual capacity and ensure safety.

The barre section (approximately 25 min) included a series of stationary exercises, with participants holding a barre for support. These exercises involved progressively increasing joint range of motion, starting with the demi and grand plié (knee bends), then progressing to battement tendu and battement glissé, (small, controlled kicks of the legs), ronds de jambe a terre (circles of the legs), battement fondu (single leg squats), adage (SLBs), and grand battement (large kicks of the legs).

Center exercises encompassed both static and dynamic movements, which were performed in the middle of the room with no external support. The center section (approximately 25 min) included port de bras (upper body and arm movements), battement tendu (small, controlled leg kicks), adage (balance training), petit allegro (small jumps), and grand allegro (large jumps, modified as necessary). The allegro exercises progressed from stationary small jumps to a series of traveling jump combinations. Dance combinations progressed in length across the intervention, with the program concluding with participants having learned specific sequences of choreography.

Statistical analysis

Deidentified data from participants who completed baseline and follow-up testing were assessed for normality using the Shapiro–Wilk’s test. Paired two-tailed t-tests were conducted for normally distributed data. All non-normally distributed data were log transformed. If the data remained non-normally distributed, a Wilcoxon signed-rank test was used. All statistical tests were conducted using IBM SPSS Statistics for Windows, version 24 (IBM Corp., Armonk, NY, USA). Statistical significance was accepted at p < .05. Per protocol, analyses were completed for all available data points, in the case of a missing datapoint, that outcome was excluded from analysis. Data were reported as mean (SD) unless otherwise specified.

Results

Participant characteristics

One hundred and one individuals responded to study advertisements and were screened for eligibility (Figure 2). Twenty-nine consecutively recruited participants met the inclusion criteria and were enrolled in the study. Seven participants were lost during follow-up due to changes in work commitments (n = 2), extended overseas holiday (n = 1), and not meeting protocol adherence requirements (and thus deemed ineligible for follow-up testing) (n = 4). A total of 22 participants completed the intervention and follow-up testing (Table 1). The mean age (SD) at baseline of this group was 56.2 (4.5) years and mean BMI (SD) was 23.9 (3.3) kg/m² (Khan et al., 1997) which was categorized as “healthy.” All included participants identified as female though all sexes were invited to participate. Of the 22 participants who completed the intervention, 17 participants (77%) were aged between 50 and 59 years, and five (23%) were aged between 60 and 69 years. There was no difference in participant characteristics at baseline between all initially enrolled participants and those who completed follow-up testing (n > .05).

Figure 2. CONSORT flow diagram for participant recruitment.

Table 1. Participant characteristics at baseline, n = 22.

	Mean (SD)	Range
Age (years)	56.2 (4.50)	50 - 69
Height (meters)	1.6 (0.06)	1.5 - 1.7
Weight (kilograms)	61 (9.30)	47.6 - 80.7
BMI (kg/m^2)	23.9 (3.30)	18.7 - 29.8
Physical activity level (IPAQ) (MET-min/week)	3071.1 (2133.80)	198 - 7771

Intervention adherence and adverse events

The intervention reported an 83% adherence rate for the 29 participants who were enrolled in the study at baseline and a 95% adherence rate for the 22 participants who completed the intervention and attended follow-up testing. Intervention adherence for the 29 participants who were initially enrolled in the study ranged from 20% to 100%. The frequency of adherence, that is the percentage of classes attended in the first 10 weeks, was reported as 78% for n = 29 and 89% for n = 22.

For the 22 participants who completed the intervention, on average, each participant was absent for 1.1 (1.1) classes. Throughout the intervention, there were 25 occurrences where classes were missed. Participants usually contacted study investigators to advise of their absence. Reasons for nonattendance included illness (five occurrences), scheduled holidays (three), unable to leave work in time (three), or other (including advised but reason not disclosed) (12). There were only two instances of “no-shows,” i.e., where no notice or reason for absence was provided.

Intervention completion was 76%, with 22 participants completing follow-up testing, out of 29 participants that were enrolled in the intervention. No adverse events were reported throughout the intervention. Further data analysis was completed for the 22 completing participants.

Questionnaires

Following the ballet intervention, no significant improvements in any questionnaire outcomes were reported (Table 2). However, more participants scored “high” levels of physical activity and less scored “low” physical activity levels at follow-up according to the IPAQ. Mean scores for the RSE at baseline (23.2 (4.9)) were within the normal range (15–25), thus indicative of good self-esteem at baseline.

Table 2. Pre- and post-intervention data for questionnaires and functional assessments.

Assessment	Outcome	Pre	Post	MD	95% CI	Change (%)	p
IPAQ	Physical activity levels	3071.1 (2133.8) L: 2^+ M: 10^+ H: 10^+	3460 (2098.7) L: 0^+ M: 10^+ H: 12^+	388.9	(−1278.2, 500.5)	12.7	0.37
SEE scale	Confidence to exercise	56.9 (19.3)	60.6 (16.1)	3.7	(−9.7, 2.2)	6.5	0.21
RSE	General self-esteem	23.2 (4.9)	24.0 (4.5)	0.8	(−2.3, 0.7)	3.7	0.30
TUG (s)	Mobility	5.0 (0.6)	4.9 (0.7)	−0.1	(−0.2, 0.4)		0.40
Five times sit-to-stand (s)	Leg strength	7.9 (1.9)	7.2 (1.9)	−0.7	(0.0, 1.3)		0.04*
Seated isometric contraction (kg)	Leg strength	25.4 (4.2)	25.5 (6.0)	0.2	(−2.6, 2.2)		0.88
Grip strength (kg)	Hand grip strength	28.6 (4.0)	28.6 (3.8)	0.0	(−1.1, 1.1)		1.00
Sit and reach (cm)	Lower body flexibility	5.5 (11.5)^	8.5 (9.9)^		0.86#		0.39
Apley back scratch (cm)	Upper body flexibility	3.3 (6.1)	2.3 (7.0)	−1.0	(0.0, 1.9)		0.05
Functional Reach (cm)	Dynamic balance	36.0 (7.1)	35.2 (5.7)	−0.8	(−2.1, 3.7)		0.57

IPAQ scores are reported as MET-min/week and categorical (L = low, M = moderate, H = high). + number of participants within each category. MD = mean difference (post—pre). Percentage change = ((post—pre)/pre) *100. *p < .05. ^median (interquartile range). # Z score.

Strength and functional assessments

Participants demonstrated an 8% faster five times sit-to-stand following the ballet intervention (p = .04) (Table 2). No other significant pre-post differences in functional assessment were obtained.

Range of motion

Left and right forward leap distance (p = .03; and p = .04, Z = −2.05, respectively) improved following the ballet intervention (Table 3). Ankle plantar flexion ROM, assessed during the calf raise movement, showed a significant decrease for the right ankle only (p = .03). There were no other differences observed between pre- and post-intervention ROM measures (Table 3).

Table 3. Pre- and post-intervention data for lower limb joint ROM during specific ballet movements.

Assessment (outcome)	Side	n	Pre	Post	MD	95% CI	p
Battement devant (hip flexion)	Left	21	66.0 (10.1)	67.1 (8.3)	1.1	(−3.6, 1.4)	0.37
	Right	22	66.7 (9.7)	68.6 (8.8)	1.9	(−5.3, 1.4)	0.24
Battement derriere (hip extension)	Left	21	18.0 (8.1)	19.7 (6.5)	1.7	(−3.9, 0.4)	0.11
	Right	22	18.7 (4.7)^^	19.4 (7.8) ^^		−1.79^#	0.07
Grand plié (knee flexion)	Left	22	99.5 (17.1)	99.1 (19.1)	−0.4	(−4.0, 4.8)	0.83
	Right	22	99.5 (18.1)	99.7 (17.9)	0.2	(−4.5, 4.1)	0.93
Demi plié (ankle dorsiflexion)	Left	22	34.9 (7.9)	34.6 (7.7)	−0.3	(−1.2, 1.8)	0.70
	Right	22	35.2 (8.4)	35.4 (6.1)	0.2	(−2.5, 2.2)	0.88
Rise (ankle plantarflexion)	Left	21	40.3 (8.7)	37.7 (8.6)	−2.6	(−0.9, 6.1,)	0.13
	Right	21	42.6 (16.9)^^	38.1 (11.2)^^		−2.23^#	0.03*
Leap (anterior displacement) (mm)	Left	18	1167.6 (189.5)	1238.0 (155.9)	70.4	(−134.2, −6.5)	0.03*
	Right	20	1120.9 (224.0)^^	1179.6 (293.6)^^		−2.05^#	0.04*
Hop (elevation) (mm)	Left	19	125.2 (34.5)	133.2 (34.2)	8.0	(−17.5, 1.5)	0.09
	Right	21	132.2 (29.2)	132.4 (28.2)	0.2	(−13.6, 13.3)	0.98
Jump (elevation) (mm)		21	230.6 (44.2)	237.9 (54.3)	7.3	(−23.7, 9.2)	0.37

*p < .05. ^{^}median (interquartile range). ^#Z score. MD = mean difference (post—pre). All data reported in degrees unless otherwise stated.

Balance assessments

A significant improvement in the CoP ellipse area (p = .04) was reported for the left leg in the parallel retiré condition following the ballet intervention (Table 4). Participants demonstrated improved balance duration on both left and right sides for all conditions (% change ranging from 6% for left leg functional eyes closed, to 22.2% for right-leg turnout), except for the right leg during the functional eyes closed condition, though these findings were not significant (Table 5). All balance variables improved for the left leg during the parallel retiré condition though similarly these findings were not significant (Table 4).

Table 4. Pre- and post-intervention balance parameters, for the four single-leg balance conditions, performed on the left leg.

Outcome	Parallel retiré					Turnout retiré					Functional eyes open					Functional eyes closed
	n	Pre	Post	95% CI	p	n	Pre	Post	95% CI	p	n	Pre	Post	95% CI	p	n	Pre	Post	95% CI	p
Duration (s)	19	22.0 (9.3)	24.2 (9.3)	(−6.9, 2.4)	0.32	18	18.8 (11.1)	22.0 (9.7)	(−7.6, 1.2)	0.14	19	23.7 (9.2)	27.1 (7.3)	(−7.7, 0.9)	0.11	19	7.9 (5.7)	8.3 (7.1)	(−3.9, 3.0)	0.78
CoP ellipse area (%IASIS^2)	19	344.2 (242.0)	223.3 (97.1)	(9.7, 232.2)	0.04*	18	217.7 (124.0)^^	219.7 (122.7)^^	−0.76^#	0.45	19	220.7 (90.2)	266.7 (374.0)	(−225.6, 133.7)	0.60	19	424.8 (220.1)	474.7 (238.6)	(−168.9, 69.2)	0.39
CoP-CoM distance M/L (%IASIS)	13	20.2 (13.5)	18.3 (11.2)	(−7.0, 10.9)	0.64	13	14.1 (10.6)^^	12.8 (5.1)^^	−0.59^#	0.55	10	10.2 (1.9)	11.5 (3.3)	(−3.0, 0.4)	0.13	15	14.6 (16.4)^^	14.96 (7.8)^^	−0.28^#	0.78
CoP-CoM distance A/P (%IASIS)	13	35.4 (47.5)	30.1 (48.8)	(−25.0, 35.6)	0.71	13	13.7 (22.9)^^	12.0 (5.3)^^	−1.01^#	0.31	10	9.1 (8.2)^^	9.6 (9.1)^^	−0.26^#	0.80	15	20.3 (30.0)^^	16.0 (26.3)^^	−0.41^#	0.68
CoP velocity M/L (%IASIS/s)	19	20.8 (4.5)	20.5 (4.5)	(−1.3, 1.9)	0.67	18	21.8 (9.4)^^	21.1 (6.5)^^	−0.68^#	0.50	19	17.6 (3.6)	17.7 (3.5)	(−1.8, 1.6)	0.93	19	25.6 (7.0)	25.6 (5.1)	(−2.8, 3.5)	0.82
CoP velocity A/P (%IASIS/s)	19	20.4 (8.3)^^	16.3 (8.4)^^	−1.01^#	0.31	18	21.0 (6.0)^^	20.5 (5.3)^^	−0.59^#	0.56	19	16.2 (8.1)^^	15.9 (5.5)^^	−0.40^#	0.69	19	30.3 (16.8)^^	27.9 (21.6)^^	−0.46^#	0.65

*significant pre to post (p < .05). ^{^}median (interquartile range). ^#Z score. Due to marker dropout, balance data was only available for a maximum of 19 participants.

Table 5. Pre- and post-intervention balance parameters, for the four single-leg balance conditions, performed on the right leg.

Outcome	Parallel retiré					Turnout retiré					Functional eyes open					Functional eyes closed
	n	Pre	Post	95% CI	p	n	Pre	Post	95% CI	p	n	Pre	Post	95% CI	p	n	Pre	Post	95% CI	p
Duration (s)	19	24.5 (7.1)	27.6 (7.7)	(−7.3, 1.0)	0.13	19	19.5 (9.8)	23.9 (9.4)	(−9.2, 0.6)	0.08	19	23.9 (9.1)	26.0 (8.0)	(−8.0, 3.9)	0.48	19	8.8 (7.9)	8.2 (7.6)	(−3.1, 4.4)	0.73
CoP ellipse area (%IASIS^2)	19	254.0 (123.3)	265.0 (154.6)	(−90.0, 67.9)	0.77	19	240.6 (78.7)	297.2 (450.8)	(−279.9, 166.7)	0.60	19	222.8 (109.6)	241.1 (190.6)	(−120.9, 84.4)	0.71	19	461.3 (233.2)	403.5 (153.9)	(−44.1, 159.7)	0.25
CoP-CoM distance M/L (%IASIS)	14	13.3 (2.6)^^	13.3 (9.3)^^	−0.68^#	0.64	11	13.4 (13.8)^^	11.9 (8.0)^^	−1.69^#	0.09	13	10.4 (9.4)^^	8.9 (1.9)^^	−1.78^#	0.08	15	15.2 (10.1)^^	13.6 (4.8)^^	−1.14^#	0.26
CoP-CoM distance A/P (%IASIS)	14	22.3 (30.9)	34.1 (45.3)	(−44.4, 20.8)	0.45	11	12.7 (13.4)^^	9.8 (4.3)^^	−1.96^#	0.05	13	10.3 (18.5)^^	7.9 (6.8)^^	−0.45^#	0.65	15	26.1 (31.7)	30.1 (40.4)	(−33.3, 25.4)	0.78
CoP velocity M/L (%IASIS/s)	19	21.2 (4.3)	22.4 (4.5)	(−2.7, 0.5)	0.15	19	20.0 (4.7)^^	20.9 (4.5)^^	−0.52^#	0.60	19	16.9 (4.6)^^	16.5 (5.3)^^	−0.93^#	0.35	19	29.6 (6.5)	29.9 (5.8)	(−4.6, 4.0)	0.89
CoP velocity A/P (%IASIS/s)	19	20.1 (5.8)^^	19.7 (8.6)^^	−0.47^#	0.49	19	22.1 (4.4)	22.9 (7.8)	(−4.7, 3.0)	0.65	19	16.3 (6.0)^^	15.7 (4.8)^^	−0.12^#	0.90	19	36.7 (16.1)	35.8 (11.1)	(−6.3, 8.0)	0.81

*significant pre to post (p < .05). ^{^}median (interquartile range). ^#Z score. Due to marker dropout, balance data was only available for a maximum of 19 participants.

Marker dropout during collection of motion analysis data prevented calculation of CoM for some participants. Therefore, a reduced sample size is reported for some balance variables.

Discussion

This single-arm study investigated a 10-week classical ballet training intervention in terms of balance, physical function, and joint ROM for adults aged 50 years and over. The findings showed improved lower limb strength, as measured by five times sit-to-stand and forward leap assessments. Improvements in balance were found in the left leg only, as measured by the CoP ellipse area in the parallel retiré balance condition. A decrease in right-ankle plantar flexion ROM was reported. High adherence rates (95% adherence for participants who completed the intervention) were reported, with no adverse events. Although this study was a single-arm design, it suggests promising results for future research wishing to evaluate the effectiveness of classical ballet training using randomized controlled designed studies.

The current study presents a safe method for challenging balance, as tasks performed during ballet classes require a reduced base of support (standing on one leg), minimal upper limb support (center floor exercises), and controlled CoM movements (weight transfers). These properties have been identified as the most efficacious training components for improving balance (Sherrington et al., 2017). The inclusion of multiple balance conditions for analyses is a strength of this study; however, significant improvements were reported in one balance condition only. This may be because the mean baseline data were similar to age predicted normative values for all static balance variations and exceeded normative ranges for dynamic balance (forward reach). Participants in the current study were perhaps too young (mean age 56.2 years) and healthy to show age-related deteriorations in balance, a finding that has been reported for adults aged 58–68 years following a Caribbean dance intervention (Federici et al., 2005). However, ballet training aimed at this age group may act to preserve physical functioning into older years as commencing regular exercise training in middle age is associated with healthy aging (Sabia et al., 2012; Sun et al., 2010).

The most effective dosage of exercise to improve balance has been identified as at least 3 hs per week (Sherrington et al., 2017). The current study did not meet these dosage recommendations, with only 1 h per week of training completed. However, the duration and class frequency of the current intervention were reflective of achievable, real-life weekly time commitments by this population. Furthermore, the average dosage of ballet training interventions for beginner dancers was previously reported as 1.6 sessions/week of 1.3 h/session (Letton et al., 2020). Interventions of other dance genres, which have reported improved physical function and balance for older adults, included at least two classes per week (McKinley et al., 2008; Sofianidis et al., 2009; Sohn et al., 2018). Thus, a greater training dosage than that utilized in the current study may be required to determine the efficacy of classical ballet training for functional and balance adaptations.

The current study showed improved lower limb strength, with quicker times obtained for the five times sit-to-stand and greater forward leap distance achieved. Although improvements in this outcome were statistically significant, they were not clinically significant. Clinically significant improvements can be determined by the minimum detectable change (MDC), which is defined as the smallest difference between repeated measurements that is not attributed to measurement error or chance. When the change between two scores is greater than the defined MDC for that test, it can be confirmed that a true adaptation has occurred (Hollman et al., 2008). The MDC for the five-fold sit-to-stand was previously reported as 2.5 s, in a study that investigated females with a mean age of 73.6 years (Goldberg et al., 2012). The mean difference between pre- and post-intervention five times sit-to-stand scores in the current study was only 0.7 s. Previously, an Argentine Tango intervention observed a 4.8-s improvement in five times sit-to-stand scores following 10 weeks of training (2-h classes, twice per week) (McKinley et al., 2008). However, frail adults were recruited in this previous study. For future ballet interventions, recruitment of older participants or those with poor balance may be needed to determine if similar functional adaptations can occur from classical ballet training.

The current study showed improvements in forward leap distance for both left and right sides. However, ankle plantar flexion ROM decreased at follow-up, while other joint ROM did not change following the intervention. The movements performed during joint ROM assessments mirrored movements which were performed during class. Hence, it was expected that improvements in these joint ranges would occur in the current study (Harris et al., 2007). Previous studies that investigated the effects of dance training on trunk and lower limb ROM have produced contradictory findings. Low-impact dance interventions (three x 60-min classes/week for 16 weeks) for females aged 60 years and over improved ROM for lower limb joints (ankle inversion, eversion and dorsiflexion, and knee extension) (Wu et al., 2016). However, a systematic review (Fernandez-Arguelles et al., 2015) concluded that there was not enough evidence to suggest that dance could significantly improve trunk and lower limb joint ROM for healthy adults aged 60 years and over. However, the systematic review found only two studies reporting on trunk and lower limb joint ROM, each investigating different dance genres: one study reported on low impact aerobic dance, performed for two × 60-min classes/week for 12 weeks (Hui et al., 2009) and the other studied ballroom dancing, consisting of one × 60-min class/week for 12 weeks (Holmerova et al., 2010). Further, only the ballroom intervention showed significant improvement in trunk and lower limb ROM. The authors cited that some aspects of these included studies (including low methodological quality, small sample size, and diversity in included studies) compounded to limit the strength of the conclusions drawn from this review. These previous findings, combined with the results for the current study (which reported both significant and insignificant changes in joint ROM) suggest that more research is required to determine if dance-based interventions can improve joint ROM in adults aged over 50 years.

Commitment to the current ballet class was positive, as participants usually contacted study investigators to advise of absences, except in two instances, which constituted only 8% of all absences. Attendance rates were strong, with a 95% attendance rate for the 22 participants who completed the intervention (and 83% for all enrolled participants). Adherence to dance-based training interventions has been compared to that of structured exercise, with most dance studies reporting adherence rates over 80% (Fong Yan et al., 2018). However, methods for the evaluation of adherence vary considerably between studies, and there is no accepted standard for the reporting of intervention adherence (Hawley-Hague et al., 2016). Thus, the current study calculated adherence in several ways (i.e., across the total number of available sessions and in 10 consecutive weeks), which all demonstrated high adherence to the current intervention. It has also been previously determined that different adherence measures should be reported based on study outcome measures (Hawley-Hague et al., 2016). To accurately determine effectiveness on health outcomes, completion and attendance should be reported in addition to duration and intensity to allow for adequate comparison to current evidence. However, duration and intensity were not reported in the current study. This may be applicable to future interventions, which might wish to categorize intervention intensities. Overall, participants thoroughly enjoyed the 10-week classical ballet training intervention. Although feedback was not formally collected as part of the intervention assessments, 10 participants informally provided testimonials after the 10-week ballet intervention, all of which were positive. In addition, many participants opted to keep attending ballet classes following the cessation of the research study.

The high levels of enjoyment reported throughout the current study indicate the potential for long-term adherence to classical ballet training.

Limitations and future directions

Single arm intervention design is a limitation of the current study. The current study was designed as a pilot investigation to determine the effectiveness of classical ballet training. Future studies should include randomized control trials in a larger cohort, as well as investigations beyond female-only populations. The social aspects of classical ballet training should also not be ignored or underestimated. The current study did not include formal, objective evaluation of cognitive outcomes, enjoyment, or motivation for attendance, however informal testimonials were provided by many participants at the end of the intervention. Thus, future studies may find it beneficial to formally report on personal motivation and reasons for attending classical ballet classes.

The intervention duration and/or class frequency during this study may not have been high enough to induce significant changes in physical function in the current healthy cohort. Intervention duration and class frequency were chosen as a realistic time commitment by this population in a real-world setting. Future studies should consider the evidence in the current dance literature, as well as the American College of Sports Medicine’s programming recommendations for training adaptations, which include at least 30 min/day of moderate intensity aerobic activity and at least 2 days/week of at least moderate intensity for resistance and flexibility activity (Chodzko-Zajko et al., 2009). Longer duration interventions, including long-term follow-ups may be beneficial, especially as exercise training is most efficacious when maintained long term.

Conclusion

Key findings from this study included significant improvements in balance and lower limb strength, including a five times sit-to-stand and forward leap, following a 10-week classic ballet intervention in adults aged 50 years and above. High adherence rates suggest classical ballet training was enjoyed by this population and may thus be maintained long term. Future studies using classical ballet interventions for middle-aged and older adults should investigate greater exercise dosage and aspects of why participants adhered well to the intervention. These findings may inform future studies of the legitimate potential of classical ballet as an effective physical activity intervention for older female adults.

Acknowledgments

The authors report no conflict of interest. This research was supported by the Australian Government Research Training Program Scholarship. The authors would like to thank Neuroscience Research Australia for providing the use of their Biomechanics Laboratory for the study.

Disclosure statement

No potential conflict of interest was reported by the authors.