Resting energy expenditure in professional dancers as an objective measure of low energy expenditure

Abstract

Objective

To investigate the prevalence of low resting energy expenditure (REE) in ballet dancers and to examine its association with low energy availability (LEA) questionnaires and physiological parameters, including menstrual function and body composition.

Methods

Participants were screened two times during a dance season. Measurements included direct REE (REE_m) via gas analysis, body composition, bone health, menstrual status and LEA questionnaires. REE_m was compared against eight predictive equations (REE_p). Metabolic suppression was defined as an REE_m/REE_p ratio <0.90.

Results

47 professional classical ballet dancers (female: n=27, age 24±5.21 years, body mass index (BMI) 18.8±1.67; male: n=20, age 25±4.27 years, BMI 23.1±1.59) volunteered. A high prevalence of suppressed REE was identified, affecting 37.5% of females and 46% of males. No single predictive equation was superior; using multiple equations improved the identification of at-risk dancers. In females, low REE was strongly associated with menstrual irregularity and a significantly later menarche. In males, low REE was linked to lower body fat%. Crucially, low REE was not reliably predicted by low body mass or BMI. Instead, the ratio of kcal/FFM kg/day was a more sensitive discriminator between low and normal REE groups in both sexes. Dancers with low REE also demonstrated less favourable bone health profiles.

Conclusion

Metabolic suppression indicative of LEA is widespread in elite dancers and is not discernible by body mass alone. Clinical screening should incorporate direct metabolic measurement, assessment of menstrual status and evaluation of body composition to effectively identify dancers at risk of the long-term health consequences of relative energy deficiency in dance.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Resting energy expenditure (REE) is closely linked to energy intake, and a low REE can serve as a marker for energy deficiency in athletic populations. Previous research has suggested that a ratio of measured to predicted REE (REE_m/REE_p) below 0.90 may indicate metabolic suppression associated with low energy availability (LEA). Menstrual irregularities in female athletes and low body fat in male athletes have been identified as potential indicators of LEA risk.

WHAT THIS STUDY ADDS

• This study demonstrates a high prevalence of metabolic suppression in elite ballet dancers (37.5% of females, 46% of males) that is not reliably predicted by low body mass or body mass index alone. The ratio of kcal per kg of fat-free mass per day (kcal/FFM kg/day) proved more sensitive than kcal/kg body weight for discriminating between dancers with low and normal REE in both sexes. While the LEAF-Q identified historical markers of LEA (later menarche and lower body fat) in females, neither questionnaire reliably detected current metabolic suppression, highlighting the need for objective physiological measurement.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Clinicians should employ multiple predictive equations when calculating REE_m/REE_p ratios, as this approach improves sensitivity for identifying dancers with metabolic suppression compared with using a single equation. Screening protocols for dancers should incorporate direct metabolic measurement alongside assessment of menstrual status in females and body composition in males, as low REE was strongly associated with these clinical markers. The trend towards lower bone mineral density in dancers with suppressed REE underscores the need for early identification and intervention to prevent long-term health consequences, including stress fractures and osteoporosis. It supports integrating REE screening into routine health monitoring in dance companies.

Introduction

With the increased emphasis on relative energy deficiency in dance (RED-D),¹ several objective methods have been suggested to identify dancers at possible risk, including resting metabolic rate/resting energy expenditure (REE). Energy homeostasis is centrally regulated, and REE is closely linked to energy intake² and contributes substantially to total energy expenditure.³ Therefore, when energy intake is insufficient, dancers are more likely to suffer suboptimal energy availability and a lower REE.⁴ Moto et al⁵ noted that menstrual disorders also lowered REE despite similar fat-free mass (FFM) and normal thyroid and reproductive hormone ranges. Mishra⁶ suggested that female athletes should consume at least 30 kcal/kg body weight (BW)/day to prevent functional amenorrhoea. Taguchi and Oshima⁷ challenged this value and provided a range of REE data for female athletes, 21.9–26 kcal/kg BW/day across various sports, including ball-game athletes and endurance runners; while for equivalent male athletes, it ranged between 23.2 and 25.7 kcal/kg BW/day. It must be recognised that these data were for Japanese athletes, and whether they can be applied to other ethnicities has not been examined in elite athletes. However, there is probably greater homogeneity than within general populations, where differences have been observed.⁸

Equations to calculate low energy availability (LEA) risk have been proposed since 2008,⁹ and Staal et al⁴ suggested a low ratio between measured. They predicted that REE (<0.9) could serve as a marker of energy deficiency. Staal et al highlighted that there is an issue with the equations used to predict REE values, especially in highly active athletes; within these equations, FFM rather than total body mass is the important parameter. The need to take into account sex,10,13 body mass index (BMI)¹⁴ and ethnicity^{15 16} is also vital, with some equations being specific to these parameters.

Another important aspect to potentially measure is the energy substrate utilisation during REE measurements. The respiratory exchange ratio (RER) provides an estimate of the amounts of fat and carbohydrate being consumed; a ratio closer to 0.8 indicates primarily fat oxidation, and one closer to 1.0, carbohydrate.¹⁷ With LEA and low body fat (energy reserves), the RER could be skewed towards increased carbohydrate use to protect the limited fat availability if energy availability worsens.¹⁸

The study aimed to examine measured REE (REE_m) of elite ballet dancers against different predictive REE (REE_p) equations and, second, to examine the RER and energy substrate utilisation within this population.

Methods

PPI statement

The ballet company has a dancer representative group that serves as a conduit to the entire company. The group was part of the methodology design and feedback mechanisms for the participants. Their guidance determined the dates on which data collection occurred and how the feedback was presented to the participants.

The dancers are contracted to train for 46 weeks a year, 36 hours/week, which includes 20 performance weeks and 26 rehearsal weeks. Volunteers were recruited at the start of the season (August 2025), screened and required to be tested at least two times during the study: once at the start of the season after their major holiday and again immediately after an intensive 4-week performance period. Fifty-five professional classical ballet dancers volunteered for the study, and 37 completed all tests; eight dropped out or had incomplete datasets.

Protocol

At each data collection period, participants were asked to arrive overnight-fasted, including no caffeinated drinks. However, they were allowed to drink water up to 30 min before their appointment. Measurements of anthropometry (height and body mass), body composition, REE and menstrual status (self-report and blood analysis) were collected. Each participant completed an LEA questionnaire according to their sex (LEAM-Q or LEAF-Q).^{19 20}

All anthropometric measurements and body composition were carried out in minimal clothing: females, leotard and tights; males, tights/shorts and t-shirts; and the scales were adjusted to account for 0.5 kg of clothing; room temperature was set at 20 °C. Stature was measured using a Secca stadiometer to the nearest 0.2 cm. Body composition, percentage body fat (BF%) and FFM were measured using a Tanita MC-780MA (Netherlands) with participants standing on the scales, feet 30 cm apart and arms straight in front of them, perpendicular to the body. The parameters recorded were height (m), body mass (kg), BF% and FFM (kg). Whole body bone health (BMD) was assessed using dual-energy X-ray absorptiometry (DXA: iDXA; GE Medical Systems, Madison, Wisconsin). Participants were supine and wore non-elasticated clothing during the scan. Bone mineral density (g/m²) and Z-score were recorded.

REE was measured by resting gas analysis using a Cosmed Metamax (Cosmed, Italy) and participant-specific facemasks. Each participant lay down on a physiotherapy plinth in a dark, isolated room and remained awake for 20 min. The lowest REE over 5 min was used as their REE_m. The recorded parameters were REE/day (kcal), relative REE (kcal/kg/day), RER and estimated daily fuel use (carbohydrate, fat and protein: g/day).

Two 5 mL venous blood samples were taken from the median cubital vein of the participant’s non-dominant arm and analysed for oestradiol (E2), luteinising hormone (LH), follicle-stimulating hormone (FSH) and prolactin (Prol) to confirm menstrual status; for males, testosterone was analysed.

Data categorisation

REE_p was calculated for each individual from a series of recognised equations: Harris-Benedict,¹⁰ Watson,¹¹ ten Haaf,²¹ Hannon,²² Owen,¹³ de Lorenzo,¹² Taguchi⁷ and Tinsley.²³

Certain parameters were categorised according to the previous literature as indicating either LEA or relative energy deficiency. REE was categorised as either normal or at risk according to the Staal et al⁴ hypothesis; individual REE_m and REE_p, and the REE_m/REE_p ratio, were calculated for each test. Scores below 90% were classified as at risk. Bone mineral density z-scores greater than zero were categorised as normal, and below zero as at risk. The at-risk body fat category was defined as <6% for male dancers and <15% for female dancers.⁴ BMI (body mass kg/height m²) scores below 18.5 kg/m² were classified as at risk.⁴ The questionnaires were analysed according to their protocols with scores: for the LEAF-Q, a total score ≥8 was considered to be at risk of LEA and female athlete triad²⁰; while for the LEAM-Q the questions on low sex drive (5F1a, 5F1b, 5F2a, 5F2b) are considered the effective self-report symptom for male LEA.¹⁹ An endocrinologist analysed blood data to confirm menstrual status and testosterone²⁴ with irregular (2–8 cycles per year), secondary amenorrhoea (females) and low testosterone (males) classified as at risk.

Statistical analysis

Data were tested for homogeneity of variance (Levene) before a series of independent t-tests on REE and anthropometric variables were carried out with sex, blood status, REE category and LEA status as independent variables. The association between the categories was analysed using χ² tests. All statistical tests were carried out using SPSS (IBM, USA), consistent with CHAMP.²⁵ Alpha level was set at 95%; p≤0.05.

Results

In total, 97 tests were carried out during the study period (10.6084/m9.figshare.30489350). Categorisation (table 1) indicated that 12 females and 4 males were potentially at risk of LEA (LEAF-Q and LEAM-Q). Eleven of the 24 female dancers were classified at risk for menstruation cycles (self-reported, confirmed by blood analysis); no males were classified as at risk (blood analysis). For BMI, 10 men and 15 females were classified as having a normal BMI. Four males and 12 females were classified as at risk for body fat. For female participants, there were no statistically significant relationships among the categories; for male participants, the only significant relationship was between the resting energy and BMI categories (p<0.05), but this should be interpreted with caution due to the low sample size.

Table 1. Frequency of participants categorised ‘at risk’ according to sex.

Sex	Resting energy		BMD		BMI		Body fat		Blood markers		LEA-Q
	Normal	At risk	Normal	At risk	Normal	At risk	Normal	At risk	Normal	At risk	Normal	At risk
Female (n=24)	15	9	20	4	15	9	12	12	11	13	12	12
Male (n=13)	7	6	13	0	10	3	9	4	13	0	9	4

BMD, body bone health; BMI, body mass index; LEA-Q, Low Energy Availability Questionnaires (LEAM-Q, Leaf-Q).

The at-risk group reported a significant 2-year later onset of menarche (table 2). As expected, males had a significantly lower body fat percentage and higher BMI (p<0.001) than females. In the BMI categories, determined by <18.5 kg/m², for the female normal menses group, 73% were classified in the normal BMI category, and for the irregular menses group, 52% had normal BMI. In the female menses subcategories, 79% of those with regular menses had body fat above 15%, while 55% of those with irregular menses had low body fat. For females, the LEAF-Q was not statistically associated with menstrual status or BMI category. There was a significant association between the LEAF-Q and REE categories (p<0.014), with those in the LEAF At Risk group more likely to be in the low REE group. There were no significant category associations for male participants.

Table 2. Participant characteristics (mean (95% CI)).

Sex (N)	Menses	Menarche (age)	Age (years)	Height (m)	Mass (kg)	BMI (kg/m2)	%BF	FFM (kg)	BMD (g/m2)	BMD Z-score
Female (n=24)	Normal (n=13)	13** (12 to 14)	24 (22 to 26)	1.65 (1.64 to 1.66)	52.5 (50.6 to 54.3)	19.1*** (18.7 to 19.9)	15.8*** (14.5 to 17.4)	43.8 (42.8 to 45.2)	1.127 (1.103 to 1.149)	0.868 (0.628 to 1.108)
	At risk (n=11)	16** (15 to 17)	23 (21 to 26)	1.66 (1.65 to 1.68)	51.7 (49.3 to 54.1)	18.5*** (17.9 to 19.3)	14.5*** (12.6 to 15.5)	43.4 (42.6 to 45.9)	1.096 (1.075 to 1.117)	0.563 (0.259 to 0.866)
Male (n=13)	All	–	25 (23 to 26)	1.78 (1.77 to 1.80)	74.4 (71.6 to 77.2)	23.2*** (22.6 to 23.9)	9.2*** (7.9 to 10.4)	67.5 (65.1 to 69.9)	1.356 (1.323 to 1.388)	1.664 (1.350 to 1.979)

Significance level: *p,0.05; **p<0.01; ***p<0.001.

BF, body fat; BMD, body bone health; BMI, body mass index; FFM, fat-free mass.

There were significant differences between the sexes for REE_m (p<0.001), kcal/day/kg BW (p<0.01), and substrate utilisation: carbohydrate (p<0.05), fat (p<0.001) and protein (p<0.001) (table 3). The RER for the whole cohort was 0.88±0.05, with no difference between the sexes or between the normal and at-risk menses groups. Several participants had suppressed REE scores (REE_M/REE_P<90%); when using the criterion of ‘at least two suppressed REE ratios’ as defined in the methods, 6 out of 13 males (46%) and 9 out of 24 females (37.5%) had suppressed REE (tables1 4).

Table 3. Anthropometric and REE data (mean±SD) for each test for REE, LEA-Q and blood status categories.

		Number of tests	First menses (years)	Age (years)	BMI (kg/m2)	%BF	BMD	BMD Z-score	REE (kcal/day)	REE (kcal/day/kg)	REE (kcal/FFM kg/day)	RER	CHO (g)	Fat (g)	Prot (g)
			Mean±SD	Mean±SD	Mean±SD	Mean±SD	Mean±SD	Mean±SD	Mean±SD	Mean±SD	Mean±SD	Mean±SD	Mean±SD	Mean±SD	Mean±SD
Resting energy expenditure category
Female	Normal	37	14±2.04**	25±5.32	18.7±1.46	15.1±3.88	1.124±0.05	0.866±0.64	1820±230***	34.4±5.47**	42.5±6.92**	0.89±0.03	276±68.0**	62±30.8	20±2.7**
	At risk	30	16±2.34**	24±5.14	18.9±1.91	14.8±4.26	1.094±0.05	0.521±0.81	1440±144***	27.7±4.34**	32.8±4.79**	0.88±0.05	211±66.6**	53±23.2	16±1.5**
Male	Normal	19		25±4.60	23.2±1.70	9.2±2.97	1.360±0.07	1.768±0.74	2210±238***	29.8±2.37**	33.0±2.56**	0.82±0.02	303±99.5	92±40.0	24±2.7**
	At risk	11		24±3.62	23.0±1.43	8.7±3.55	1.345±0.09	1.444±0.94	1739±433***	27.1±2.57**	25.5±5.51**	0.88±0.05	252±64.5	67±43.6	19±5.3**
LEA questionnaire category
Female	Normal	27	14±1.65**	25±3.83	18.8±1.38	16.3±3.42*	1.099±0.05	0.635±0.59	1697±244	32.7±5.99	39.9±6.39	0.88±0.03	256±77.0	60±20.4	19±2.8
	At risk	35	16±2.47**	24±6.07	18.7±1.89	14.0±4.22*	1.115±0.06	0.730±0.82	1589±284	29.9±5.75	36.2±8.29	0.88±0.04	235±72.3	55±32.1	17±3.2
Male	Normal	23		26±4.45	23.1±1.60	9.3±3.28	1.348±0.08	1.629±0.83	2122±325	29.2±2.63	31.7±4.32*	0.83±0.19	290±95.0	89±38.0	23±3.6*
	At risk	7		22±2.69	23.1±1.65	7.9±2.55	1.377±0.08	1.771±0.78	1733±476	27.5±2.85	25.2±5.52*	0.89±0.06	25 972.8	63±52.6	19±5.9*
Blood status (menstruation and testosterone)
Female	Normal	30	13±1.98**	24±4.83	19.1±1.34	15.7±3.91	1.126±0.05*	0.868±0.58	1769±285***	32.7±4.53	40.5±7.45**	0.88±0.03	255±63.9	66±31.0	19±3.4**
	At risk	37	16±2.02**	24±5.57	18.5±1.86	14.1±4.07	1.096±0.05*	0.563±0.84	1541±217***	30.0±6.64	35.9±7.37**	0.89±0.04	237±81.3	52±23.5	17±2.2**
Male	Normal	30		25±4.27	23.1±1.58	9.04±3.15	1.355±0.08	1.664±0.81	2025±397	28.8±2.75	30.1±5.36	0.85±0.17	283±89.7	82±42.6	22±4.7

Significance level: *p,0.05; **p<0.01; ***p<0.001.

BF, body fat; BMD, body bone health; BMI, body mass index; CHO, carbohydrate; FFM, fat-free mass; REE, resting energy expenditure.

Table 4. Physiological parameters for the low and normal REE groups for male and female dancers (mean (95% CI)).

	REE category	BF (%)	FFM (kg)	kcal/BMkg/day	Kcal/FFM kg/day	BMI (kg/m2)	BMD (g/m2)	BMD Z-score
Females (n=9)	At risk	14.6 (13.0 to 16.2)	44.6 (42.9 to 46.2)	27.7 (25.9 to 29.4)	32.7* (30.9 to 34.6)	19.0 (18.2 to 19.8)	1.095 (1.073 to 1.141)	0.52 (0.20 to 0.84)
(n=15)	Normal	14.9 (13.2 to 16.5)	44.0 (42.4 to 45.6)	33.2 (31.2 to 35.2)	41.8* (38.9 to 44.8)	19.1 (18.4 to 19.7)	1.118 (1.097 to 1.141)	0.81 (0.56 to 1.06)
Males (n=6)	At risk	9.2 (6.3 to 12.0)	68.9 (64.9 to 72.9)	27.4 (25.4 to 29.3)	26.2* (21.6 to 30.8)	23.2 (22.1 to 24.4)	1.345 (1.272 to 1.418)	1.44 (0.72 to 2.17)
(n=7)	Normal	9.6 (8.1 to 11.1)	67.1 (63.6 to 70.5)	29.9 (28.7 to 31.1)	33.0* (31.7 to 34.3)	23.4 (22.5 to 24.3)	1.360 (1.319 to 1.403)	1.80 (1.42 to 2.18)

Significance level: *p,0.05.

BF, body fat; BMD, body bone health; BMI, body mass index; FFM, fat-free mass; REE, resting energy expenditure.

Seven male dancers with suppressed REE had multiple <90% ratios across the different predictive REE equations (table 3); four dancers scored below 90% for all equations, one for six equations, and two for four equations. There was no significant difference between the male at-risk and normal REE groups for age, BMI, bone health, body fat % or FMM. For age, there was a trend for dancers with suppressed REE to be younger. Dancers with a suppressed REE showed a tendency to have a low body fat % (<5%) (4 out of 7), but a notable minority were classified as normal body fat %. For those classified as having normal REE, the trend is much more obvious, with 82% having scores above 5% body fat. Interestingly, the suppressed REE group had a slightly higher, though physiologically non-significant, FMM than the normal REE group, which was unexpected (table 5). The categorisation of REE is not reflected in bone health data, with the suppressed REE group reporting slightly lower BMD and Z-scores than the normal REE group (table 5). The low REE group had significantly lower relative REE for both body mass (kcal/BW kg/day) and FMM (kcal/FFM kg/day) than the normal REE group (p<0.01).

Table 5. Comparison of measured versus predicted REE (REE_m/REE_p) (mean (95% CI)) using different equations.

Males	REEm	REEp
		Tanita	Cunningham	H-B	ten Haaf	Hannon	de Lorenzo	Tinsley
kcal/day	2025 (1871 to 2179)	1964 (1894 to 2034)	1995 (1939 to 2050)	1814 (1768 to 1861)	2032 (1974 to 2089)	2070 (2041 to 2097)	1906 (1862 to 1946)	2043 (1979 to 2109)
% REEm		106 (85 to 109)	105 (84 to 108)	114 (84 to 108)	103 (94 to 112)	101 (81 to 104)	108 (88 to 113)	103 (82 to 105)
N≤90%		4/13	4/13	3/13	4/13	4/13	4/13	4/13

Females	REEm	REEp
		Tanita	Cunningham	H-B	ten Haaf	Hannon	Owen	Tinsley	Watson	Taguchi
Normal menses
kcal/day	1769 (1654 to 1885)	1326 (1293 to 1359)	1463 (1440 to 1486)	1350 (1327 to 1374)	1481 (1458 to 1505)	1801 (1790 to 1813)	1033 (1009 to 1058)	1418 (1392 to 1445)	1806 (1778 to 1833)	1214 (1186 to 1241)
% REEm		134 (97 to 136)	121 (88 to 122)	132 (95 to 133)	120 (87 to 121)	98 (71 to 99)	149 (124 to 174)	108 (91 to 126)	84 (77 to 89)	126 (115 to 137)
N<90%		0/13	0/13	0/13	0/13	0/13	0/13	0/13	0/13	0/13
Disordered menses
kcal/day	1541 (1467 to 1615)	1318 (1271 to 1366)	1455 (1418 to 1492)	1340 (1310 to 1371)	1473 (1435 to 1511)	1796 (1778 to 1816)	1024 (985 to 1064)	1408 (1365 to 1452)	1795 (1766 to 1824)	1204 (1159 to 1249)
% REEm		118 (105 to 125)	106 (95 to 112)	115 (103 to 121)	105 (94 to 111)	86 (77 to 90)	148 (135 to 161)	107 (98 to 116)	83 (71 to 98)	125 (115 to 137)
N<90%		1/11	1/11	0/11	1/11	7/11	0/11	1/11	7/11	1/11

The ‘N≤90%’ row indicates the number of participants in this subgroup whose REE_m/REE_p ratio fell below 0.90 for that specific predictive equation. For the classification of a participant as ‘At risk REE’ in other tables, a combined criterion (eg, low ratio on ≥2 equations) was used.

REE, resting energy expenditure; REEm, Measured resting energy expenditure; REEp, Predicted resting energy expenditure.

Within the female group, those with at least two suppressed REE ratios were classified as having low REE ratios and were more likely to have irregular menses (63%) than dancers with normal REE ratios, who were reported to have regular menstrual cycles (87%) (table 5). The Watson and Hannon predictive equations captured the majority of the disordered menses group, but they also flagged five dancers with regular menses (table 3). A significant difference was noted between the regular and irregular menses groups for measured REE (p<0.01). As reported for the male dancers, the use of numerous REE_p equations to predict potential issues seems beneficial, as all female dancers, except one, reported at least 2 REE_m/REE_p ratios below 90%. The categorised at-risk REE group (table 5) started menses significantly later than the normal REE group by nearly 2 years (p<0.01). They also had slightly higher FMM and BMI, which is in contrast to the male dancers. However, bone health parameters were lower for this group. Replicating the male dancers, the at-risk REE group had significantly lower kcal/day FFM/kg and kcal/day BW/kg than the normal REE group (p<0.01).

Discussion

The present study of 47 professional dancers reveals a high (47%) prevalence of RED-D,¹ characterised by a suppressed metabolism that is strongly linked to menstrual dysfunction in females and occurs independently of BMI in both sexes. These data support previous research by Staal et al⁴ and Doyle-Lucas et al²⁶ who both reported suppressed REE, either compared with predictive equations or against age-matched controls. The cohorts for the three studies were similar in age, height, body mass and BF%, and all participants were recruited from elite national ballet companies. Still, the REE_m in the present study was only comparable to Staal et al (females: 1504 kcal/day, males: 1967 kcal/day), with Lucas-Doyle et al ’s data being 150 kcal/day lower (females: 1367 kcal/day). The RER, as an indicator of fuel utilisation, is similar between Doyle-Lucas et al and the current study’s female dancers with normal menses. Still, neither was as high as the irregular menses group. Although not statistically significant, a trend was observed suggesting an inverse relationship between BF% and RER. If confirmed, this could indicate a metabolic strategy to preserve limited fat stores by increasing carbohydrate oxidation, a phenomenon observed in other athlete populations under LEA (table 1). This has implications for fuelling strategies for dancers, which would require more frequent meals to maintain energy levels and prevent fatigue, especially in the later stages of the training day.²⁷

There was widespread metabolic suppression (low REE) across the cohort, with the suggested Staal et al⁴ 90% ratio between REE_M/REE_P accounting for 37.5% of females and 46% of males. None of the predictive equations was superior using Staal’s ratio, but using multiple equations was beneficial for identifying at-risk dancers, as most exhibited low REE across several calculations. The online supplemental data (10.6084/m9.figshare.30489350) show that, among female dancers, low REE status was not a fixed characteristic, with 80% of dancers classified as low REE also having at least one test with a normal REE ratio. Sixty-three per cent of this cohort also had irregular menses and menarche at 15+ years old, 3 years later than the normal REE group, which is a strong indicator of long-term energy deficiency.²⁸ Golden and Carlson²⁹ reported that it was LEA rather than exercise load that suppresses the gonadotrophin-releasing hormone and values below 30 kcal/FFM kg/day were associated with low LH and FSH. This clinical indicator of LEA in females³⁰ was observed in both male and female cohorts, with kcal/FFM kg/day being more sensitive than kcal/BW kg/day in differentiating between the REE groups. This potentially highlights the importance of monitoring FMM and body fat% with this population, which is often considered to be controversial due to pressures on body thinness and eating disorders.³¹

Our findings reveal a paradoxical relationship between REE and body composition in dancers, as determined by REE_m/REE_p ratios below 90%. Contrary to expectations, male dancers categorised as having low REE exhibited slightly higher FFM but lower BMI compared with those with normal REE. This finding aligns with research by Trexler et al,³² who reported that metabolic adaptation to energy deficiency can occur independently of changes in body mass or composition. In the male low-REE cohort, the group had a lower body fat percentage, and we propose that REE is suppressed to protect the remaining body fat in case energy availability declines.

In female dancers, those categorised as having low REE showed slightly higher FFM and BMI, which contrasts with typical expectations. Also, the body fat percentage is similar between the low and normal REE groups. Still, the group with irregular menses, accounting for 63% of the low-REE group, had a lower body fat percentage. Detrimental bone health is the result of long-term REDs,³³ and the current low-REE cohort was beginning to manifest the effects of LEA on bone health, as demonstrated by lower, but not significant, BMD and z-scores in both male and female cohorts. This paradox may be explained by the concept of ‘adaptive thermogenesis’ described by Müller and Bosy-Westphal,³⁴ where the body reduces energy expenditure beyond what would be predicted by changes in body composition alone.

The relationship between self-reported symptoms and objective markers of LEA in this cohort was complex and sex-specific. The LEAF-Q successfully identified female dancers with two key historical markers of long-term energy deficiency (later menarche and lower body fat percentage).^{28 31} Yet, it failed to differentiate based on current metabolic suppression (REE_m/REE_p ratio) or bone health. This disconnect suggests that in elite dance populations, where body image pressures and symptom normalisation are high,³¹ questionnaires may be prone to under-reporting or may capture different aspects of the REDs spectrum than a physiological measure like REE. Specifically, the LEAF-Q may be more sensitive to chronic, historical LEA rather than acute metabolic adaptations. Alternatively, the non-specific nature of some questionnaire items may limit their usefulness, as gastrointestinal issues and injuries are prevalent among dancers regardless of energy status. In contrast, the LEAM-Q showed no significant associations with any physiological or metabolic variable in male dancers. This finding aligns with the developers’ own work,¹⁹ which noted challenges in questionnaire validation and suggested that specific symptom questions (eg, low libido) may be more effective indicators than a composite score. The very low number of male dancers classified as ‘at risk’ by the LEAM-Q (n=4) in our study also limits statistical power and underscores the ongoing difficulty in developing practical, self-report screening tools for LEA in males. Therefore, while questionnaires remain a useful first-step screening tool for raising awareness, they should not be considered a substitute for direct metabolic measurement when available. Further research is needed to determine the predictive validity of these tools relative to objective markers such as REE in elite dancer populations.

Clinical implications

The findings of this study have several practical implications for clinicians and practitioners working with elite dancers. Direct measurement of REE, combined with the REE_m/REE_p ratio using multiple predictive equations, provides a valuable non-invasive indicator of LEA. The use of a battery of predictive equations (rather than reliance on a single equation) significantly improves sensitivity for identifying dancers with metabolic suppression, as no single equation proved superior in this cohort. Clinicians should therefore calculate REE_m/REE_p ratios using at least 4–5 established equations and classify dancers as ‘at risk’ if they fall below 0.90 on two or more equations. The REE_m data and graphical output from metabolic testing offer an educational opportunity to discuss energy demands with dancers. Presenting the calculated energy deficit can help dancers understand the gap between their current energy intake and their physiological requirements. This objective feedback may be more impactful than dietary advice alone, particularly in populations where body image pressures can hinder open discussion of eating behaviours.

Clinicians should be aware that the markers associated with low REE differ by sex. In female dancers, menstrual status is a critical indicator. Those with irregular menses or later menarche (≥15 years) should be prioritised for REE screening, as they are significantly more likely to exhibit metabolic suppression. In male dancers, very low body fat percentage (<6%) emerged as a key risk factor, even in the absence of low BMI. Body composition assessment should therefore be an integral part of routine screening for male dancers. The finding that kcal/day FFM/kg was a more sensitive discriminator between low and normal REE groups than kcal/day BW/kg underscores the importance of regular body composition monitoring. FMM should be used as the denominator when calculating relative energy expenditure, and practitioners should advocate for including bioelectrical impedance analysis or similar methods in routine health screening, despite sensitivities around body composition testing in dance populations.

While dancers in this study reported that the REE test was neither invasive nor tiring, the overnight fasting requirement poses logistical challenges given their demanding schedules. To facilitate compliance, practitioners should schedule testing early in the morning, before training commences; provide on-site facilities (fridge, microwave, kettle) to allow dancers to eat immediately after testing; and offer clear written instructions and explain the rationale for fasting to improve adherence.

The trend towards lower BMD and Z-scores in dancers with low REE, while not statistically significant in this cohort, reinforces the need for proactive bone health monitoring. Dancers identified as having suppressed REE should be prioritised for DXA scanning and considered for early nutritional intervention to mitigate long-term consequences, including stress fracture risk and osteoporosis.

Limitations

The participant group comprised only one professional ballet company in the UK, and therefore, numbers were limited but similar to those in previous published research on dance.^{4 26} Even though 16 Global Majority dancers were in the company, only five volunteered for the study, which was not enough to form a subcategory. Purposeful recruitment of this population would allow better insight into whether different equations and/or categorisation are needed, or whether the selection/audition process for entry into a ballet company makes this moot. Further information on eating disorders/disordered eating from questionnaires^{19 20} was not collected, which would have provided a further dimension to the study.

Conclusion

This study underscores the importance of using direct metabolic measurement, menstrual status and body composition to screen for energy deficiency in dancers, as low REE was not reliably associated with body mass or BMI in this cohort. Instead, low REE was strongly linked to menstrual irregularities in females and low body fat in males, with both sexes demonstrating less favourable bone health profiles. The use of REE_m/REE_p ratios below 90% as a predictor of low REE is strengthened when multiple predictive equations are employed. While self-report questionnaires like the LEAF-Q and LEAM-Q can provide useful contextual information regarding historical symptoms, they appear less sensitive for detecting current metabolic suppression than objective REE measurement. A multifaceted assessment approach, combining subjective screening with objective physiological markers, is therefore essential for early identification of dancers at risk and for preventing the long-term health consequences of RED-D.

Data availability statement

Data are available in a public, open access repository.