Project FURTHER: A Physiological, Biomechanical, and Psychosocial Profile of an All-Female 6-Day Ultramarathon Race

Abstract

Background

Ultramarathon performance reflects a complex interplay of physiological, biomechanical, and psychosocial factors, yet data from multi-day events remain limited, particularly in females. The demands and key determinants of performance in these events, including the longest ratified world-record event, the 6-day ultramarathon, are not well-characterized. This study aimed to provide a comprehensive, multi-disciplinary profile of the demands and performance determinants in an all-female 6-day ultramarathon.

Methods

10 heterogenous female athletes (27–48y, Tier 2–5, V̇O_{2 peak}: 21.0–62.9 ml∙min^–1·kg^–1) ran a self-paced, certified 6-day ultramarathon. Data were collected over the preceding year, 2–4d pre-race, during the race, and post-race. We assessed V̇O_{2 peak} with ventilatory thresholds and running economy; haemoglobin mass; body composition; total daily energy expenditure (TDEE, via doubly-labelled water) and dietary energy intake (EI); continuous heart rate, glucose, and core temperature; neuromuscular and cognitive fatigue; biomechanical parameters; in-race blood sampling to assess inflammatory markers; and psychosocial outcomes.

Results

All athletes achieved personal distance records over 6 days (463.6±199.6 km), with the top athlete covering 901.8 km, breaking the female world record. Exercise intensity was low (~46±6%V̇O_{2 peak}), but demands were extreme: up to 20 h·day⁻¹ racing with 82,699±17,192 steps·day⁻¹, total sleep 5.2±1.9 h·day⁻¹, 4188±1109 kcal·day⁻¹ EI, and 6698±2071 kcal·day⁻¹ TDEE; including three of the highest female TDEEs reported. Average active race pace was 8:41±2:27 min:sec·km⁻¹ and the average decline in running velocity from the first 2 days to the last 2 days was 9±13.1%. There was no evidence of neuromuscular or cognitive fatigue 24 h post-race. The top athlete, who ran 258.2 km farther than the next finisher, had similar lab-tested physiology compared to the next four top finishers, but spent the most time on course with the highest TDEE, EI, running and total step count, and exhibited significantly greater neuromuscular fatigue, inflammatory markers, sleep deprivation, and pacing decline. The only clear difference between the top athlete and the next 4 athletes was years of running experience (29 vs. 9.7±7.4 years).

Conclusions

The 6-day ultramarathon imposes an extreme physical and psychological challenge. Success may depend less on physical capacity and biomechanical efficiency than on the ability to persevere through fatigue.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40798-026-01041-w.

Key Points

• This field-based descriptive study implemented an integrative, multi-disciplinary framework to characterize the physiological, psychosocial, and biomechanical demands of a 6-day ultramarathon in female athletes.

• Athletes experienced an extreme physiological challenge as evidenced by the exceptionally high energy expenditure, sleep deprivation, circulating inflammatory biomarkers, and reported physical pain which necessitated coping strategies to reframe this psychosocial collective experience.

• World class 6-day ultramarathon performance may depend less on laboratory-derived physiological capacity and biomechanical efficiency than on the ability to tolerate, regulate, and persist through escalating fatigue, discomfort, and sleep disruption over multiple days.

• These results indicate that determinants of multiday ultramarathon performance differs from traditional endurance predictors, and reinforces the importance of female-specific research to inform training, nutrition, and recovery practices.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40798-026-01041-w.

Introduction

Ultramarathon is a rapidly growing sport, with rising participation fueling greater competition and advancing insight into the limits of human endurance. Over the past 20 years, the number of races has surged from fewer than 1000 in 2005 to more than 7000 in 2019, with total finishers rising from 89,597 (18% female) to 682,156 (23% female) in the same period [1]. Ultramarathon races are defined by distance or time, including any race longer than the traditional marathon distance (> 42.2 km) or exceeding 6 h [2–4]. The International Trail Running Association further classifies these events into ultra-M (42–69 km), ultra-L (70–99 km), and ultra-XL (≥ 100 km) [5]. Typical ultramarathon distances include 50 km, 50 miles, 100 km, and 100 miles [6]. The longest officially ratified world-record race is the 6-day ultramarathon, a fixed duration event in which runners aim to accumulate as much distance as possible over 6 days [7]. This race traces its roots back to the late 1700s, originating as a pedestrianism event, and is often regarded as a foundational precursor to modern ultrarunning [8].

Performance success at the ultramarathon distance is thought to be influenced by more than the typical physiological determinants of shorter distance running. These ultramarathon-specific success factors include diverse and varied terrain, environmental conditions, pacing techniques, nutrition strategies, gastrointestinal symptom mitigation, sleep deprivation, isolation, and a greater possibility for both fall and overuse injuries [9, 10]. Berger et al. summarized the key health and performance determinants of ultramarathon as thermoregulation, oxygen uptake, running economy and biomechanics, injury and fatigue, digestive function, and nutritional and psychological strategies [1]. Recently, Tiller and Millet (2025) proposed that muscle damage and neuromuscular fatigue are the primary impediments to performance. However, most prior research has focused on ultramarathon distances ≤ 100 miles or durations ≤ 24 h [11]. While multi-day mountain ultramarathons such as the Tor des Géants (~ 330 km with 150h time limit) and Ultra-Trail du Mont-Blanc (UTMB; ~ 170 km) have received increasing scientific attention, research specifically addressing races lasting ≥ 6 days remains scarce [12–15] and the determinants and limitations of performance in these extreme settings are not well-understood. Furthermore, there are very few studies across the field of sport science that include data on world-record performances [16, 17]. Finally, as with many areas of sports science, there is also a lack of female-specific research, despite the widely held belief that female athletes may be physiologically and psychologically suited for prolonged endurance exercise [18, 19].

Project FURTHER was a certified 6-day, self-paced ultramarathon featuring 10 female athletes with diverse training histories—ranging from the best athlete in the world at the time of the event to recreational runners with less than 1 year of experience [20]. This project aimed to explore the boundaries of human endurance and to facilitate a female attempting to break both the female 6-day world record (883.631 km), as well as the male world record (1036.800 km) at the time of the event. This was a performance-optimized, officially ratified competition, where research was carefully integrated. Considerations were made to optimize performance including careful selection of a flat looped course, climate conditions, as well as medical, nutritional, and sponsored equipment provisions including customized clothing and certified racing shoes. To support comprehensive, yet minimally invasive, field-based data collection, a robust research infrastructure was implemented. However, all aspects of race strategy including pacing, rest and recovery, and nutrition/hydration were ultimately at the discretion of each athlete and their support crew [21]. With its unique race conditions, diverse athlete profiles, and extensive research infrastructure, the FURTHER race provided a rare opportunity to explore the determinants to 6-day ultramarathon performance in females.

The purpose of this field-based descriptive study was to implement a multi-disciplinary approach to characterize the physiological, psychological, and biomechanical demands of a 6-day ultramarathon in females. A case-comparison approach was used, focusing on the athlete who covered the greatest distance and the next four elite or highly trained finishers to explore potential determinants of top performance over a 6-day ultramarathon. This capstone article serves as a comprehensive overview of the race and the research methods used prior to the publication of several discipline specific companion papers to follow.

Methods

Experimental Overview

This longitudinal within-subject repeated-measures, mixed-methods, descriptive study tracked n = 10 female ultramarathon runners across an 11-month period (health screening and V̇O_{2 peak} testing beginning in April 2023), culminating with a prospective research study conducted during a certified 6-day ultramarathon race (March 6th to 12th, 2024).

The pre-race testing period was conducted 2–4 day pre-event. Seven athletes completed the submaximal graded treadmill test after the neuromuscular and cognitive fatigue tests: n = 6 on day -4 and n = 1 on day -2. Due to delayed athlete arrivals caused by weather, n = 4 athletes completed the fatigue tests on day -2 after the fasted venous blood draws, haemoglobin mass (Hbmass), and DXA body composition scan (Table 1).Three days prior to the race start, participants ingested DLW for the measurement of pre-race and in-race total daily energy expenditure (TDEE).

Table 1.

Timing of experimental procedures

	Pre-testing	Post-testing
	2–4 days prior to race	1-day post-race
Submaximal exercise test	✓ (− 2 to 4 days)
Body composition (DXA)	✓ (− 2 days)
Blood collection	✓ (− 2 days)	✓
Haemoglobin mass	✓ (− 2 days)
Neuromuscular fatigue testing	✓ (− 2 to 3 days)	✓
Cognitive fatigue testing	✓ (− 2 to 3 days)	✓
DLW urine collection	✓ (− 3 days)	✓
Body mass	✓ (− 2 to 4 days)	✓
Interviews	✓ (− 1 to 3 months)	✓ (+ 1 month)

	In-race testing
	Daily monitoring (1–2 times/day)	Continuous monitoring	Periodic monitoring
Blood collection	✓
DLW and urine collection	✓
Body mass	✓
Dietary intake		✓
GPS + HR watch		✓
Continuous glucose monitoring		✓
Core temperature monitoring		✓
Oura smart ring		✓
Inertial measurement units		✓
Force plate			✓
Markerless motion capture			✓
Psychosocial observation			✓

DXA dual X-ray absorptiometry, DLW doubly-labelled water, GPS global positioning system, HR heart rate

The in-race testing was conducted by 28 researchers organized into discipline specific sub-teams and included daily assessments of dietary intake (DI), body mass, and appetite questionnaires, as well as the collection of blood and urine samples (Table 1). Wearable technology was used to monitor running speed, heart rate (HR), interstitial fluid glucose concentration, core temperature, sleep/recovery cycles, and accelerometery. Psychosocial observations were performed throughout the race, and qualitative interviews (two pre-race and one post-race) were conducted to examine the athletes’ experiences with stressors and pain while training prior to the event, and during the race. These qualitative findings were used to contextualize and interpret quantitative results on fatigue and performance.

Post-race testing took place the day following the race (Table 1). Unless specified, we implemented best practice sport field-based research methods as recently outlined by our research group [21].

Participants

The athletes (age: 38 ± 7 years) were from various sporting backgrounds. Eight of the 10 athletes reported a previous running training history [10.3 ± 9.7 years (range: 1—29) of experience], with four athletes being world class or elite ultramarathoners [15 ± 10.8 years (range: 6—29)] (Table 2). Two athletes had previous careers in competitive team sports (volleyball and basketball). The 10 athletes were performance classified for ultramarathon running [22] for the year prior to the event as follows: n = 1 world class (Tier 5), n = 3 elite/international level (Tier 4), n = 1 highly trained/national level (Tier 3), and n = 5 trained/developmental (Tier 2). One athlete was using hormonal contraceptives, n = 8 were pre-menopausal, and n = 2 were perimenopausal based on self-reported menstrual cycle history. Athletes provided verbal and written informed consent prior to testing. This study was approved by the University of British Columbia clinical ethics review board (#H23-03110 and #H23-02179) in accordance with the Declaration of Helsinki, except for registration in a database.

Table 2.

Athlete characteristics

	All athletes	Top 5 finishers		Bottom 5 finishers
	Athletes #1–10	Athlete #1	Athletes #2–5	Athletes #6–10
Age (years) (n = 10)	37.8 ± 6.2 (27.3–48.4)	42.2	36.7 ± 8.0 (27.3–44.7)	37.7 ± 5.7 (32.6–48.4)
Height (cm) (n = 10)	171 ± 8 (157–182)	175	172 ± 11 (157–182)	169 ± 7 (157–180)
Body mass (kg) (n = 10)	71.9 ± 25.8 (47.6–135.2)	59.0	59.4 ± 6.2 (52.2–66.7)	84.5 ± 31.5 (47.6–135.2)
Body mass index (kg/m2) (n = 10)	24.7 ± 9.1 (18.8–48.3)	19.2	20.1 ± 1.0 (18.8–21.3)	29.5 ± 10.9 (19.3–48.3)
Fat mass (kg) (n = 10)	18.8 ± 17.0 (6.2–61.8)	9.6	10.4 ± 3.3 (6.2–14.1)	27.4 ± 20.6 (10.0–61.8)
Fat-free mass (kg) (n = 10)	53.5 ± 9.2 (40.4–68.2)	50.3	50.5 ± 7.4 (42.1–59.5)	56.6 ± 10.8 (40.4–68.2)
Body fat (%) (n = 10)	22.8 ± 11.1 (10.5–47.5)	16.0	17.3 ± 5.8 (10.5–22.9)	28.6 ± 12.7 (17.6–47.5)
tHb mass (g·dL⁻1) (n = 7)	835 ± 130 (639–1066)	884	781 ± 124 (639–864) n = 3	872 ± 190 (685–1066) n = 3
Relative tHb mass (g·kg⁻1) (n = 7)	12.4 ± 2.7 (6.4–14.9)	14.6	13.0 ± 1.6 (12.0–14.9) n = 3	11.0 ± 4.0 (6.4–13.7) n = 3
Total blood volume (ml) (n = 7)	6302 ± 1227 (4337–7807)	6654	5914 ± 1631 (4337–7595) n = 3	6573 ± 1486 (4923–7807) n = 3
Relative blood volume (ml·kg⁻1) (n = 7)	93 ± 23 (52–131)	110	99 ± 28 (82–131) n = 3	82 ± 26 (52–99) n = 3
V̇O2 peak (ml·min⁻1·kg⁻1), (n = 10)	45.7 ± 13.8 (21.0–62.9)	59.1	55.5 ± 8.4 (45.3–62.9)	35.1 ± 10.8* (21.0–49.5)
GET (% of V̇O2 peak), (n = 10)	66.2 ± 6.6 (51.1–76.8)	74.9	68.0 ± 5.9 (64.1–76.8)	63.0 ± 6.9 (51.1–68.8)
RCP (% of V̇O2 peak), (n = 10)	85.7 ± 3.8 (77.3–91.1)	90.2	84.1 ± 4.9 (77.3–89.1)	86.1 ± 3.3 (82.7–91.1)
Running economy (kcal·km⁻1·kg⁻1), n = 10)	0.95 ± 0.07 (0.82–1.08)	0.99	0.97 ± 0.06 (0.92–1.05)	0.93 ± 0.09 (0.82–1.08)
Running experience (years), (n = 8)	10.3 ± 9.1 (1.0–29.0)	29	9.7 ± 7.4 (4.0–18.0) n = 3	6.0 ± 6.4 (1.0–15.0) n = 4
Run training duration (h·week⁻1), (n = 8)	14.3 ± 3.3 (8.0–18.0)	18	13.0 ± 0.0 (13.0–13.0) n = 3	14.3 ± 4.8 (8.0–18.0) n = 4
Total training duration (h·week⁻1), (n = 8)	20.4 ± 5.3 (11.0–26.0)	21	17.7 ± 2.9 (16.0–21.0) n = 3	22.3 ± 7.5 (11.0–26.0) n = 4
50 km PR (h), (n = 6)	6.05 ± 3.21 (3.35–11.28)	3.35	6.27 ± 4.35 (3.45–11.28) n = 3	7.07 ± 3.79 (4.38–9.75) n = 2
100 km PR (h), (n = 6)	15.83 ± 9.05 (7.14–30.21)	7.14	10.54 ± 3.29 (8.03–14.27) n = 3	28.11 ± 2.98* (26.00–30.21) n = 2
100 mile PR (h), (n = 4)	14.74 ± 1.89 (12.69–17.82)	12.69	16.10 ± 2.43 (14.39–17.82) n = 2	14.07 n = 1
Farthest race/Run (Prior to the event)† (km), (n = 10)	176.18 ± 120.65 (50.00–434.52)	434.52	225.07 ± 79.02 (160.93–326.01)	85.39 ± 48.28* (50.00–168.98)

Data as Mean ± SD and range (min–max). *p<0.05 comparing Athletes #1–5 to Athletes #6–10. Body mass was measured via scale, whereas fat mass, fat-free mass, and body fat percentage were derived by dual-energy X-ray absorptiometry. GET gas exchange threshold. PR, personal records prior to FURTHER (as recorded in an official sanctioned race event of any type/surface. RCP respiratory compensation point. ^†within a single dedicated competition or effort, with the same 6-day competitions rules implement in this race/study. Total training duration includes cross-training and/or strength conditioning, along with running specific training

Race Characteristics and Rules

The 4.099 km (2.5469 mile) course (~ 80% crushed gravel / ~ 20% paved) looped Lake Cahuilla in Palm Springs, CA, USA, with minimal elevation change (a single slope of 1.21 m over 107 m; ~ 1.1% grade). The race started at 09:05 am March 6th, 2024, and finished at 10:05 am March 12th, 2024 (1 h time change due to daylight savings). The race was run in accordance with the International Association of Ultramarathoners (IAU Label #20241006) and USA Track and Field (USATF Sanction #149006) rules and regulations. The course route, as well as all time- or distance-based achievements were officially measured according to the Association of International Marathons and Distance Races standards (AIMS certification # CA2400DT). Lap times were recorded using electronic timing (Mylaps ProChips affixed to shoes), with timing mats capturing both half- and full-lap splits. Per IAU and USATF regulations pacing was prohibited, athletes wore World Athletics-approved shoes, and athletes were subject to USADA testing. Environmental conditions were recorded every 5 min using an on-site weather station.

The site featured a primary pit-stop area containing individual athlete tents, which was accessible on every loop and supported by approximately 5–8 crew members per athlete. A second aid station midway around the course provided fluids and sports nutrition products tailored to the athlete’s self-reported preferences (e.g., electrolyte beverages, gels, and bars). An adjacent athlete village was established with individual trailers equipped for extended rest and sleep at the athletes’ discretion. Transport between the course and village was provided by golf carts when needed. Adjacent to the course, a chef-led sports nutrition team provided catering and integrated dietary assessment. The area also featured recovery modalities, a research tent, and full medical support from three rotating physicians specialized in emergency and sports medicine.

Testing Protocols and Equipment

Graded Exercise Testing

Graded exercise testing was conducted on a motorized treadmill with continuous collection of expired gases and ventilation (TrueOne 2400, Parvo Medics, Salt Lake City, USA) ~ 11 (n = 9) and 5–7 (n = 7) months prior to the race, with a submaximal test 4 days before the race (n = 7). Peak oxygen uptake (V̇O_{2 peak}) was determined as the highest value between the two most recent tests performed within the 11 months prior to the race. V̇O_{2 peak} is denoted rather than V̇O_2max as some athletes did not push to exertional failure, although all achieved ≥ 95% age-predicted HRmax and peak respiratory exchange ratios > 1.03, which is the proposed maximal effort criterion for trained runners (mean 1.09 ± 0.05) [23]. Athletes were instructed to follow their normal pre-competition dietary routine and avoid strenuous exercise in the 48 h before each test.

The tests began at 1% grade with an initial speed of 4–6 km·h⁻¹, corresponding to individualized comfortable walking pace. The protocol initially consisted of 3 min stages, with speed increasing by 1 km·h⁻¹ per stage, and participants having 30 s rest periods to collect capillary (fingertip) blood lactate (EDGE, ApexBio, Taiwan, China). For the submaximal test performed in the pre-competition period 4 days prior to the race, testing was terminated once 4 mmol·L⁻¹ of blood lactate was attained. For the other tests, once blood lactate concentrations reached 4 mmol·L⁻¹, the protocol transitioned to 1 min stages increasing by 1 km·h⁻¹ or 1% grade (individual preference) without breaks until volitional withdrawal or V̇O_2max.

The gas exchange threshold (GET) and respiratory compensation point (RCP) were determined using an online application (https://www.exercisethresholds.com/) based on the criteria outlined by Keir et al. [24]. Two investigators independently identified the V̇O₂ corresponding to the GET and RCP. Running economy was determined as kcal·km⁻¹·kg⁻¹ in the last minute of the first running stage (7–10 km·h⁻¹). GET, RCP, and running economy were determined from the submaximal test 4 days prior to the race in as many athletes as possible (n = 7), and from the graded exercise test with the highest V̇O_{2 peak} in the remaining athletes.

Body Composition

Body composition was assessed 2 days prior to the race using DXA (GE Healthcare, Brøndby, Denmark) according to best practice methods [25].

Haemoglobin Mass (Hbmass)

The optimized carbon monoxide (CO) rebreathing method was used to assess haemoglobin (Hb) concentration, hematocrit (Hct), blood volume (BV), red cell volume (RCV), plasma volume (PV), and Hbmass prior to the race. Upon arrival to the laboratory in a hydrated and fasted state, participants rested in a seated position for ≥ 15 min prior to measurement. They then performed a CO-rebreathe dosed at 1.0 ml·kg⁻¹ for 2 min according to methods by Schmidt & Prommer (2005) [26]. Samples were analysed immediately using an on-site blood gas analyser (ABL90 FLEX, Radiometer, Denmark). Due to issues with equipment and athlete availability (delayed athlete arrivals due to weather), results are only available for seven athletes.

Neuromuscular Fatigue

Using established methods (e.g., [27],), torque data were recorded from the dorsiflexors of the dominant lower limb (determined subjectively as the ‘leg used to kick a ball’) 2–4 days before the event and ~ 24 h after the race. Brief (~ 2 sec) isometric maximal voluntary contractions (MVCs) were performed to record maximal strength, with a single electric stimulus delivered to the common fibular nerve during and ~ 1 sec after each MVC to quantify voluntary activation (VA [28];). Dorsiflexors were chosen as the muscle group of choice as they contract eccentrically in the contact phase, and thus would be expected to experience damage over the course of the event [29]. One athlete opted out of stimulation, so VA data were collected from n = 9 participants.

Cognitive Fatigue

Cognitive fatigue was assessed with three digital tasks 2–4 days before and ~ 24 h after the race: go/no-go visual reaction time [30], Corsi block-tapping [31], and trail-making [32]. For the go/no-go task, 20 unmeasured practice trials were presented before 150 measured trials, with 20% being no-go trials. For the Corsi block-tapping task, the sequence length started at three and increased by one provided the athlete correctly replicated the pattern for at least one of two trials. Five practice trials were completed before the measured trials began. Finally, for the trail-making task, athletes completed five practice and 10 measured trials for version A (all numbers) and B (alternating numbers and letters). One athlete opted out of post-race cognitive testing; therefore, the sample size was 10 pre-race and 9 post-race.

Muscle Damage/Inflammatory Markers

Venous blood samples were collected from the antecubital vein 2 days prior the race (n = 10; fasted), immediately post-race (< 1 h of finishing, day 6; n = 10), and 24h post-race (n = 9; fasted). Daily in-race collection was also attempted where possible (race days 1–5; n = 3 to n = 6 depending on race day). Due to the practical constraints of the race, the daily in-race blood samples were not standardized for exercise, caffeine, or nutrition. However, all in-race blood samples occurred within a daily 5-h window (between ~ 7:00 and 12:00), with the relative timing consistent between days within athletes.

Samples were sent to a commercial laboratory for analysis of creatine kinase (CK) and C-reactive protein (CRP) (Quest Diagnostics, CA, USA) or processed and stored in liquid N₂ prior to transfer to -80 ºC for later analyses, using standard best practice [33]. EDTA samples were centrifuged at 2000 g for 10 min at 21ºC. Following 1 h of coagulation, SST samples were centrifuged at 2000 g for 15 min at 21ºC. EDTA plasma samples were analysed for cardiovascular and inflammatory protein biomarkers using Proximity Extension Assay (Olink, Sweden, Target 96 Cardiovascular II).

Dietary Intake (DI)

Ad libitum dietary intake (DI) was assessed for 1–3 days before, throughout the 6-day race, and for 1 day after. During the race, meals were prepared and weighed by an on-site chef and kitchen staff, with additional options for individual requests/external foods and self-preparation from athletes. Food and beverages were weighed with digital scales and recorded using paper forms or a SmartSheet on iPads with pre-populated items, supplemented by food photographs. A team of six trained researchers with nutrition expertise were available 24 h/day to assist athlete crews with DI assessment using gold-standard methods [21]. Dietary data were entered into Food Processor software (version 11.14.9; Trustwell, OR, USA) to analyse energy and macronutrient intake, with each entry cross-verified for accuracy by a second researcher.

Doubly-Labelled Water (DLW)

Energy expenditure was measured via DLW throughout the race. The DLW dose was between 80 and 180 g of 10% enriched H₂¹⁸O and 4–9 g of 99.8% enriched ²H₂O (Sigma-Aldrich) for a range of 47.6–131.5 kg of body mass. Total body water was estimated as 73% of fat-free mass (estimated as 70% of body mass). Athletes provided six urine samples at the following timepoints during FURTHER: 1) 2–3 days prior to race start in the morning prior to the DLW dose; 2) 4–6 h after the DLW dose; 3) within 4 h prior to the race start (day 0); 4) race day 3; 5) race day 6, as close to the race finish time as possible (− 5 to + 2 h from race finish); and 6) 3-day post-race.

Isotope enrichments were measured via OA-ICOS (Off-Axis Integrated Cavity Output Spectroscopy) at Calorify laboratory in Scotts Valley, CA, with laboratory methods coefficient of variation of 2.1%. The natural logarithms of ²H and ¹⁸O were plotted against time, with the slopes of the regression lines representing the elimination rates of ²H and ¹⁸O (k_H and k_O, respectively). Dilution spaces for ²H (N_H) and ¹⁸O (N_O) were determined by dividing the isotope enrichments by the intercepts of these regression lines. Total body water and CO₂ were estimated using the methods describe by Speakman et al. [34], with TDEE calculated using the slope-intercept method [35]. The food quotient was assumed to be 0.86, in line with current practices [35] and representative of a typical diet on a population level [36]. Quality control (QC) measures to screen for erroneous data included post-processing QC checks for standard R² > 0.999 and that the duplicate measurements are within 1–2 per mil for ²H and 0.1–0.2 per mil for ¹⁸O (depending which sample: baseline, enriched, or final). Final QC checks include a dilution space ratio between 1.00 and 1.07 and that the final samples are > 8 per mil above the sample baseline [37].

Continuous In-Race Measurements

Racing Velocity and Heart Rate

Racing velocity, pace, and HR were determined using GPS watches (n = 7, Coros Apex 2 Pro; Coros Wearables Inc., Irvine, California, USA; n = 3, Garmin, personal sponsorship varied models; Garmin Ltd. Olanthe, Kansas, USA) with HR determined by optical armband or chest-strap telemetry. Watches provided back-up optical wrist-based HR monitoring if the straps were disconnected. Athlete #1 had a malfunction in data transfer following the race, and all watch data were lost. As such, HR was not recorded for this individual, and racing velocity was determined using race timing mats. Athlete #5 only had 2 days of GPS watch data, and therefore her HR was also removed from the reported data.

Continuous Glucose Monitoring (CGM)

Athletes were instrumented with CGM sensors (Supersapiens, Abbott Libre Sense, Illinois, USA) over the triceps brachii 1–3 days prior to the race. The sensors recorded minute-by-minute interstitial fluid glucose concentrations over a 14-day window. Sensors were replaced mid-race if they were no longer transmitting data (n = 4). Athletes wore small Bluetooth-enabled Supersapiens ‘Energy Bands’ that allowed for storage of data (2 min bins) when phones, which could not be carried on course, were out of range. Athlete #4 was missing 1 day of active glucose concentrations; therefore, results are presented as n = 9.

Core Body Temperature

Telemetric core temperature pills (e-Celsius Performance, BodyCap, France; sampling frequency every 60 sec) were ingested ~ 30 min prior to the start of the race, and athletes ingested ~ 1 new pill per day throughout the race in anticipation of sensor excretion. Individual data were inspected, and new sensor data were included once the following criteria were met: (1) The core temperature sensor remained active internally for at least 2 h following ingestion, (2) temperature readings were above 36.5 °C, and (3) temperature readings did not fluctuate > 1 °C within a 5-min period [38, 39]. When multiple sensors were recording data in the same individual at the same time, the average temperature from all the sensors was used once the new sensor met the criteria above and had stable readings (> 5 min) within 0.25 °C of the other sensor(s) [40]. Missing data were linearly interpolated when time periods were ≤ 5 min in duration, otherwise missing data were excluded. Final data for each athlete were smoothed with a Gaussian weighted-average function (MATLAB R2024a, MathWorks Inc., Natick, MA, USA) [41]. Issues with data transfer resulted in limited core temperature readings on day 1. One athlete declined pills after day 3, and athlete #7 was missing 1 day of active core temperature readings; therefore, results are presented as n = 8.

Sleep/Recovery Cycles

Nine athletes wore smart rings (Oura Ring Gen 4, Oura Ltd, Oulu, Finland) for the duration of the race. The rings detect sleep periods using a combination of accelerometery, skin temperature, HR, heart rate variability, mathematical modelling of the circadian rhythm, and machine learning [42, 43], with 94–96% sensitivity to detect sleep compared to polysomnography [43, 44].

Periodic In-Race Measurements

Inertial Measurement Units (IMUs)

Athletes wore custom-made garments outfitted with a pocket over the sacrum (at the level of S2) to house a small (~ 5 g) inertial measurement unit (IMU; Plantiga Technologies, Vancouver, Canada; 8-h recording period with triaxial accelerometer (± 32G) and gyroscope, recording at 416 Hz) at all times on course. The sacral IMU position was chosen, because vertical displacement of the sacrum is strongly correlated (r = 0.95) with vertical displacement of the center of mass during running [45]. IMUs were swapped out at intervals throughout the race to be recharged and to upload data. As a first step for data processing, IMU signal timestamps were cross-referenced with GPS and activity log data to identify periods of activity and rest. During periods of activity, vertical acceleration signals were re-oriented using a custom R script (V.4.4.2; R Development Core Team) to align with gravity to account for differing sensor orientations and filtered using a 5Hz low-pass 4th order Butterworth filter ([46, 47]; R Core Team, 2024). Periods of running were defined as steps that included a flight time (both feet off the ground) and were identified using the CARL classification algorithm [48]. During running, initial contact and toe-off events were identified using the zero-crossing points of the vertical acceleration. During walking, local minima were used to define initial contact for each step. Biomechanical outcomes were calculated for the dominant activity-type (walking or running) in non-overlapping 60-s windows.

Psychosocial Interviews and Observations

Seven athletes participated in this part of the study, which involved (1) ethnographic participant observations [49], and (2) a series of three semi-structured qualitative interviews [50] about their stress and pain-related experiences leading up to, and during, the race. Seven months pre-race, three research team members observed and interacted with the athletes at a 5-day training camp to get to know them and their athletic backgrounds to inform the semi-structured interview guides. Athletes then took part in a series of two interviews 5- and 1-month pre-race, where they were asked about their athletic histories, and experiences with stressors, pain, and injury during training leading up to the race. Informed by these interview data, fieldnotes from ethnographic observations were taken during the 6-day race, focusing on how the athletes were coping with stressors and experiencing pain throughout the race in situ. A final retrospective interview was conducted with each athlete 1-month post-race to explore how they were making sense of stress and pain-related processes post-event. Data were analyzed using narrative thematic analysis [51]; we identified patterns within the athletes’ stories and our field notes of how they coped with stressors and pain over time in relation to the 6-day ultramarathon.

The researchers engaged in reflexivity throughout the research process, identifying how our positioning and subjectivities as sport psychology researchers, mental performance consultants, former athletes, and relationality with the athletes during the observations and interviews shaped the data collection, analysis, and findings presentation. In this paper, we report on the overarching findings pertaining to how athletes coped with stressors and pain. In companion papers, we will engage in in-depth exploration of different facets of these pain-related coping processes with a focus on (1) how athletes negotiated cultural sport narratives of ‘pushing through pain’ to story their experiences of pain leading up to, during, and after the race, and (2) how gendered narratives surrounding femininity and the body shaped the women’s coping processes in relation to pain.

Questionnaires

Athletes (n = 8) completed a modified Low Energy Availability in Females Questionnaire (LEAF-Q) with retrospective recall approximately 10-weeks post-race to provide information on athlete training habits (within the past 2 years prior to the race, menstrual cycle, dietary practices related to disordered eating (within the past 28 days prior to the race), general/gut health, illness, and injury (within the past 1–2 years prior to the race).

Data Analysis

Data presented herein are primarily descriptive owing to the small sample size, and athletes are ordered relative to the distance covered in the race (i.e., Athlete #1 covered the farthest distance, while Athlete #10 covered the least distance). Where possible, we descriptively compared the top finisher to the next 4 athletes who completed the most distance and were of similar performance caliber in order to explore what sets a World-Class (Tier 5) athlete apart from national/international (Tier 3&4) athletes in this event. As some of the athletes who completed the least distance were not ultramarathon runners prior to training for this event, their data were used to characterize the demands of the race, but not to identify factors and determinants of world-class performance. Specific data processing and analysis procedures for each test/measurement are detailed in the respective sections above. Unless specified, missing data were excluded from the analysis and only the individuals with complete data sets are presented. We give the sample size for each variable for transparency. For continuous biomechanical and physiological variables including running velocity, pace, HR, core temperature, interstitial fluid glucose, and accelerometry, only time spent moving (running and walking), while on-course was included in the data analysis to focus solely on the physiological demands of active competition. Time was classified as follows: days as 9:00 am–8:59 am; Morning 6:00 am–11:59 am; Afternoon 12:00 pm–17:59 pm; Night 18:00 pm–5:59 am. The active and inactive time periods were determined using the timing mats and corroborated with GPS watches and smart rings. Time spent off-course was largely passive (e.g., golf-carted to trailer, or resting at the athlete pit-stop area/athlete village off-course). Time spent racing in the moderate intensity domain below GET, heavy intensity domain above the GET but below RCP, and severe intensity domain above RCP, were determined via the HR associated with the GET and RCP from the graded exercise tests [52].

Statistical Analysis

Statistical analyses were performed using Prism 10 (GraphPad, San Diego, California, USA) and R (Version 4.4.2, R Core Team, 2022). Data were assessed for normality with Shapiro–Wilk tests. We compared the top 5 vs. bottom 5 finishers using independent two-tailed Student’s t tests. Student’s paired one-tailed t tests were used to assess changes in pre–post data. Repeated-measures ANOVA were used to examine outcomes across race days #1–6, with Tukey corrections for post hoc analysis. For IMU-derived biomechanical outcomes, we conducted analyses with and without running velocity as a covariate. Associations between variables were assessed with Pearson’s correlations. For protein biomarkers, p < 0.05 was used as the level of significance after adjusting for false discovery rate (FDR) using the Benjamini–Hochberg procedure. Each day of the race is presented as a 24 h segment beginning at 9:00 am on March 6, 2024. Data are presented as mean ± standard deviation (SD) and range (min–max), with alpha set at p < 0.05.

Results

All athletes completed the 6-day race, though the proportion of active race time and rest varied amongst individuals. Athlete #1 covered the greatest distance (901.754 km), which was 258.2 km farther than the next finisher (Fig. 1) and broke the women’s world record of 883.631 km (Sandra Barwick in 2005). All athletes achieved their farthest distance to date. One athlete (Athlete #10, Tier 4 athlete) had several injuries impacting her race preparation and, therefore, did not push to her physical limit, but did still run her personal farthest distance and is included in the mean data (Fig. 1).

Fig. 1.

Total distance ran and total daily energy expenditure (TDEE) for each athlete over 6 days of continuous racing

Athlete Characteristics

There was a wide range of physical and traininghistory characteristics across participating athletes. The top athlete (Athlete #1) did not differ from the next four finishers in most physiological and demographic characteristics, except for greater running experience, higher weekly training volume and personal record times across various race distances (Table 2). Athlete #1 had the third highest V̇O_{2 peak}, which was similar to the next 4 finishers (59.1 ml·min⁻1·kg⁻1 vs. 55.5 ± 8.4 ml·min⁻1·kg⁻1). When examining classic determinants of performance [17], Athlete #1 had a slightly poorer running economy than the next 4 finishers (0.99 vs. 0.97 ± 0.06 kcal·km⁻¹·kg⁻¹), but the highest %V̇O_{2 peak} at RCP (90 vs. 84 ± 5%). Other physical characteristics were not overtly different from the next 4 top athletes including body composition outcomes (body mass, fat-free mass, body fat percentage) and age. Athlete #1 did differ from the next 4 finishers in years of running experience (29 vs. 9.7 ± 7.4 years) and in their farthest race distance prior to this event (434.52 vs. 225.07 ± 79.02 km). Athletes #1–5 differed from those that completed the least distance (Athletes #6–10) in: V̇O_{2 peak} and previous race performances (100 km PR and farthest race/run), but not statistically in: age, body composition, Hbmass, blood volume, ventilatory thresholds, running economy, years of running experience, or weekly training duration (Table 2).

Race Performance

Athletes had total active racing durations between 22.7 (15.8% of the possible 144 h) to 102.7 h (71.3%) over 6 days. Active duration was not different between days (Table 4). Athletes ran on average 77.3 ± 33.3 km per day, ranging from 30.1 to 150.3 km per day (Table 3; Fig. 2A). Athletes covered the greatest distance on day 1 (Table 4), including Athlete #1 who covered the greatest daily distance overall (213.1 km). During the event, Athlete #1 broke 12 women’s world records (WR): 300-mile, 400-mile, 500-mile, 500-km, 600-km, 700-km, 800-km, 900-km, 3-day, 4-day, 5-day, and 6-day WRs (Table S 1).

Table 4.

Daily race performance and recovery metrics across a 6-day ultramarathon

	Day 1	Day 2	Day 3	Day 4	Day 5	Day 6
Distance (km) (n = 10)	113.9 ± 59.0 (32.8–213.1)	76.2 ± 44.1* (24.6–184.4)	68.9 ± 41.0* (4.1–151.7)	62.7 ± 30.6* (16.4–139.4)	68.9 ± 31.6* (16.4–118.9)	73.0 ± 29.9* (32.8–110.7)
Active Duration (h) (n = 10)	13.9 ± 4.4 (4.9–19.7)	10.8 ± 3.9 (2.9–19.3)	10.2 ± 5.0 (0.7–18.3)	10.0 ± 4.3 (1.8–17.6)	10.8 ± 3.8 (2.4–15.3)	11.5 ± 4.5 (3.0–20.1)
Time active morning (h) (n = 10)	4.4 ± 1.4 (1.4–5.9)	3.7 ± 1.3 (1.0–4.9)	3.4 ± 1.8 (0.7–5.8)	4.3 ± 1.8 (0–5.9)	3.6 ± 1.1 (1.7–5.2)	3.6 ± 1.2 (1.2–5.4)
Time active afternoon (h) (n = 10)	4.1 ± 1.7 (0–5.6)	2.9 ± 1.3 (0.9–5.44)	2.6 ± 1.7* (0–5.12)	2.8 ± 1.1* (1.4–4.9)	3.3 ± 1.7 (0–5.6)	3.2 ± 1.3 (1.2–5.3)
Time active night (h) (n = 10)	5.3 ± 2.2 (3.2–8.9)	4.2 ± 2.6 (1.0–8.9)	4.2 ± 2.9 (0–8.4)	2.9 ± 2.7* (0–8)	4.0 ± 2.3 (0–6.6)	4.7 ± 3.2 (0–10.8)
Active pace (min:sec·km−1) (n = 10)	7:45 ± 2:23 (5:29–12:36)	8:41 ± 2:34 (6:16–14:08)	9:09 ± 2:16 (6:24–14:23)	9:06 ± 2:36 (6:41–15:11)	9:34 ± 2:10* (7:06–13:44)	9:01 ± 2:50 (5:28–14:11)
Fastest lap velocity (kph) (n = 10)	10.0 ± 2.3 (5.7–12.7)	8.9 ± 2.2 (5.2–12.2)	8.5 ± 2.0 (4.9–11.9)	8.3 ± 2.2* (5.2–12.7)	8.2 ± 2.0* (4.9–11.9)	8.5 ± 2.5 (4.7–12.1)
Slowest lap velocity (kph) (n = 10)	6.3 ± 1.8 (4.2–9.6)	5.2 ± 1.5 (3.5–9.0)	5.6 ± 1.4 (3.4–7.6)	5.1 ± 1.3 (2.9–8.1)	4.5 ± 0.7* (3.7–5.8)	4.8 ± 2.0 (2.7–9.7)
Running step count (# steps) (n = 10)	85,627 ± 73,330 (1,184–200,899)	51,899 ± 63,436 557–195,578)	41,938 ± 49,022* (0–146,898)	29,032 ± 33,976* (727–117,558)	30,590 ± 32,898* (964–102,735)	38,131 ± 43,844* (241–116,527)
Total step count (# steps) (n = 10)	111,907 ± 56,492 (36,104–203,266)	87,086 ± 48,935 (26,664–196,690)	71,508 ± 47,462* (12–154,830)	66,605 ± 28,297* (18,319–119,917)	69,511 ± 29,449* (10,644–109,381)	89,574 ± 36,940 (43,451–137,772)
Mean flight time (sec) (n = 6)	0.074 ± 0.015 (0.045–0.087)	0.071 ± 0.017 (0.042–0.089)	0.073 ± 0.015 (0.048–0.092)	0.075 ± 0.010 (0.062–0.086)	0.073 ± 0.012 (0.050–0.084)	0.078 ± 0.012 (0.056–0.090)
Mean stance time (sec) (n = 6)	0.263 ± 0.014 (0.249–0.290)	0.266 ± 0.015 (0.243–0.290)	0.266 ± 0.016 (0.250–0.287)	0.265 ± 0.015 (0.246–0.290)	0.269 ± 0.017 (0.251–0.292)	0.262 ± 0.018 (0.244–0.288)
Mean vertical GRF impulse (BW·sec−1) (n = 6)	0.351 ± 0.016 (0.334–0.373)	0.353 ± 0.025 (0.324–0.379)	0.354 ± 0.022 (0.326–0.381)	0.354 ± 0.022 (0.328–0.385)	0.357 ± 0.016 (0.344–0.386)	0.363 ± 0.023 (0.328–0.39)
Mean vertical GRF peak (BW) (n = 6)	2.276 ± 0.119 (2.041–2.362)	2.261 ± 0.143 (2.054–2.433)	2.248 ± 0.128 (2.081–2.435)	2.252 ± 0.115 (2.134–2.411)	2.241 ± 0.109 (2.114–2.371)	2.310 ± 0.151 (2.120–2.482)
Mean step rate (steps·min−1) (n = 6)	175 ± 7 (164–183)	174 ± 9 (164–186)	175 ± 7 (166–185)	176 ± 8 (164–184)	174 ± 8 (163–184)	175 ± 8 (165–186)
Active heart rate (bpm) (n = 8)	132 ± 12 (114–148)	120 ± 16 (95–136)	122 ± 15 (98–138)	121 ± 15 (99–142)	115 ± 9* (104–134)	117 ± 21* (94–154)
Active time in moderate intensity (%) (n = 8)	69.6 ± 15.9 (51.1–99.5)	89.1 ± 14.1 (57.1–100)	91.0 ± 12.7* (62.3–100)	84.9 ± 20.1 (46.7–100)	93.6 ± 8.7* (74.9–100)	90.0 ± 23.5 (31.9–100)
Active time in heavy intensity (%) (n = 8)	29.4 ± 14.8 (0.5–44.7)	10.7 ± 13.7 (0–41.8)	8.6 ± 12.3* (0–36.5)	14.0 ± 19.4 (0–53.3)	6.2 ± 8.2* (0–23.5)	10.0 ± 23.5 (0–68.1)
Active time in severe intensity (%) (n = 8)	1.0 ± 1.5 (0–4.2)	0.2 ± 0.4 (0–1.1)	0.4 ± 0.6 (0–1.3)	1.1 ± 1.9 (0–5.3)	0.2 ± 0.6 (0–1.6)	0 ± 0 (0–0)
Active Core Temp (℃) (n = 8)	37.6 ± 0.4 (37.1–38.4)	37.6 ± 0.2 (37.4–38.0)	37.7 ± 0.4 (36.9–38.1)	37.7 ± 0.3 (37.3–38.1)	37.7 ± 0.3 (37.3–38.2)	37.6 ± 0.3 (37.2–38.0)
Active continuous glucose (mmol·L −1) (n = 9)	6.3 ± 0.3 (5.8–6.8)	6.7 ± 0.8 (5.9–8.0)	6.6 ± 1.0 (5.5–8.5)	6.9 ± 0.6 (6.1–8.0)	6.5 ± 0.9 (4.9–7.7)	6.6 ± 0.9 (5.5–8.1)
Sleep (h) (n = 6)	4.8 ± 2.2 (1.1–7.3)	5.5 ± 2.6 (0.5–7.2)	5.3 ± 2.1 (1.6–7.4)	4.2 ± 2.6 (0.9–6.4)	5.7 ± 2.2 (1.5–7.4)	5.7 ± 1.7 (3.4–8.0)

Data as Mean ± SD and (range). *p<0.05 from day 1. Days as 9:00 am–8:59 am. Morning 6:00 am–11:59 am; Afternoon 12:00 pm–17:59 pm; Night 18:00 pm–5:59 am. Time in intensity domains determined via HR monitoring, with moderate intensity domain occurring below GET, heavy intensity domain occurring between GET and RCP, and severe intensity domain occurring above RCP

GRF ground reaction force, BW bodyweights, SPM steps per minute. Active time excludes off-course rest periods

Table 3.

Race performance and recovery metrics across a 6-day ultramarathon

	All athletes	Top 5 finishers		Bottom 5 finishers
	Athletes #1–10	Athlete #1	Athletes #2–5	Athletes #6–10
Distance (km·day−1) (n = 10)	77.3 ± 33.3 (30.1–150.3)	150.3	90.0 ± 11.5 (84.0–107.3)	52.5 ± 20.0* (30.1–80.6)
Total active duration (h) (n = 10)	67.1 ± 20.8 (22.7–102.7)	102.7	71.2 ± 8.9 (60.2–82.1)	56.8 ± 23.5 (22.7–85.9)
Percent time active morning (%) (n = 10)	36.9 ± 10.7 (17.1–51.8)	26.5	35.7 ± 9.0 22.7–42.3	40.0 ± 13.6 (17.1–51.8)
Percent time active afternoon (%) (n = 10)	27.7 ± 5.1 (17.5–36.0)	27.9	28.1 ± 6.0 (23.2–36.0)	27.5 ± 6.2 (17.5–32.2)
Percent time active night (%) (n = 10)	35.3 ± 11.4 (21.2–53.7)	45.7	36.3 ± 13.7 (22.8–53.7)	32.5 ± 12.1 (21.2–50.8)
Active pace (min:sec·km−1) (n = 10)	8:41 ± 2:27 (6:38–13:46)	6:50	7:49 ± 1:24 (6:38–9:46)	10:14 ± 2:44 (7:32–13:46)
Total pace (min:sec·km−1) (n = 10)	22:55 ± 11:52 (9:35–47:54)	9:35	16:10 ± 1:50 (13:25–17:08)	30:58 ± 11:59* (17:52 ± 47:54)
Total running step count (# steps) (n = 10)	277,216 ± 273,997 (4,104–860,439)	860,439	370,367 ± 186,312 (247,875–647,320)	86,051 ± 59,070* (4,104–149,490)
Total step count (# steps walking and running) (n = 10)	496,191 ± 208,014 (196,625–898,618)	898,618	566,820 ± 118,251 (427,120–689,275)	359,202 ± 133,929* (196,625–552,761)
Running step proportion (%) (n = 10)	48.9 ± 32.7 (1.4–95.8)	95.8	64.8 ± 23.7 (39.0–93.9)	26.8 ± 24.0* (1.4–63.3)
Mean flight time (sec) (n = 6)	0.074 ± 0.013 (0.042–0.092)	0.075 ± 0.003 (0.073–0.080)	0.069 ± 0.017 (0.042–0.092)	0.080 ± 0.004* (0.075–0.090)
Mean stance time (sec) (n = 6)	0.265 ± 0.015 (0.243–0.292)	0.280 ± 0.014 (0.258–0.292)	0.266 ± 0.014 (0.247–0.290)	0.257 ± 0.010* (0.243–0.272)
Mean vertical GRF impulse (BW·sec−1) (n = 6)	0.355 ± 0.020 (0.324–0.390)	0.379 ± 0.009 (0.363–0.386)	0.347 ± 0.016 (0.324–0.372)	0.355 ± 0.021 (0.328–0.390)
Mean vertical GRF peak (BW) (n = 6)	2.27 ± 0.12 (2.04–2.48)	2.18 ± 0.08 (2.11–2.30)	2.26 ± 0.14 (2.04–2.48)	2.31 ± 0.08 (2.18–2.43)
Mean step rate (steps·min−1) (n = 6)	175 ± 7 (163–186)	168 ± 5 (163–176)	175 ± 7 (164–186)	178 ± 6 (170–186)
Active heart rate (bpm) (n = 8)	121 ± 12 (106–137)	-	124 ± 16 (106–137)	120 ± 11 (108–135)
Active time in moderate intensity (%) (n = 8)	86.4 ± 10.4 (68.1–98.3)	-	95.9 ± 2.4 (93.5–98.3)	80.6 ± 8.8* (68.1–90.7)
Active time in heavy intensity (%) (n = 8)	13.1 ± 10.2 (1.3–31.6)	-	3.9 ± 2.6 (1.3–6.5)	18.7 ± 8.7* (9.3–31.6)
Active time in severe intensity (%) (n = 8)	0.5 ± 0.5 (0–1.3)	-	0.2 ± 0.3 (0–0.5)	0.7 ± 0.6 (0–1.3)
Active core temperature (℃) (n = 8)	37.7 ± 0.2 (37.4–37.8)	37.8	37.6 ± 0.2 (37.5–37.8)	37.7 ± 0.2 (37.4–37.8)
Active continuous glucose (mmol·L −1) (n = 9)	6.6 ± 0.6 (5.9–7.4)	6.9	6.8 ± 0.8 (5.9–7.4)	6.3 ± 0.4 (5.9–6.8)
Total sleep (h·day−1) (n = 6)	5.2 ± 1.9 (1.5–6.9)	1.5	5.6 ± 0.1 (5.5–5.7)	6.2 ± 0.6 (5.8–6.9)

Data as Mean ± SD and range (min–max). *p<0.05 comparing Athletes #1–5 to Athletes #6–10. Days as 9:00AM–8:59AM. Morning 6:00am–11:59am; Afternoon 12:00 pm–17:59 pm; Night 18:00 pm–5:59am. Time in intensity domains determined via HR monitoring, with moderate intensity domain occurring below GET, heavy intensity domain occurring between GET and RCP, and severe intensity domain occurring above RCP. GRF ground reaction force, BW, bodyweights. Active time excludes off-course rest periods

Fig. 2.

Individual in-race active distance (A), duration (B), racing velocity (C), sleep hours (D)*, total step count (E), and running step count (F) across 6 days of ultramarathon running. *Values of 0 h of sleep indicate missing data. All athletes slept minimally ~0.5 h/day

In-Race Physiology

Active race pace over the 6 days ranged from 6:38 min:sec·km⁻¹ to 13:46 min:sec·km⁻¹ (average 8:41 ± 2:27 min:sec·km⁻¹; Table 3; Figs. 2 and 3). Athlete #1 maintained an average race pace of 6:50 min:sec·km⁻¹, while the athlete who completed the second farthest distance maintained a faster average pace (6:38 min:sec·km⁻¹) over a shorter distance. The average decline in running velocity from the first 2 days to the last 2 days was 9 ± 13.1%, with Athlete #1 having the largest decline (26.2%), the next 4 athletes having a decline of 13.4 ± 9.9%, and the last 5 athletes having 5.8 ± 16.1% (p = 0.3 between Athletes #1–5 and #6–10). Athletes spent more time racing during the day (64.7 ± 11.4%) compared to overnight (35.3 ± 11.4%; p = 0.002). Athlete #1 spent 54.3% of racing time in the day and 45.7% overnight, which was more relative night running than the next 4 athletes who spent 63.7 ± 13.7% racing in the day and 36.3 ± 13.7% overnight. Athletes #6–10 spent 67.5 ± 12.1% racing in the day and 32.5 ± 12.1% overnight which was not different from Athletes #1–5 (p = 0.5). Athletes #2 and #4 both took a light day mid-race due to symptoms of overexertion and/or injury but were able to continue racing the following day (Fig. 2A).

Fig. 3.

Active racing velocity (n = 10), heart rate (n = 8), core temperature (n = 8), ambient temperature, and humidity across 6 days of racing. Data as Mean (black) and SD (grey). AM, morning 6:00 am–11:59 am; PM, afternoon 12:00 pm–17:59 pm; NT, night 18:00 pm–5:59 am. DST, daylight savings time. Physiological data includes active racing time; ambient temperature and humidity include full race data (active and recovery). Any break in the data indicates there were zero athletes on the course.

Racing HR (n = 8) was on average 121 ± 12 bpm, ranging from 106 to 137 bpm across athletes and 63.7 ± 4.9% of HRmax. This equated to 86.4 ± 10.4% time spent in the moderate intensity HR domain, 13.1 ± 10.2% time spent in the heavy intensity HR domain, and 0.5 ± 0.5% time spent in the severe intensity HR domain across the 6 days. The top 3 athletes (not including Athlete’s #1 and 5 whose HR data were incomplete) spent greater proportion of time in the moderate intensity domain (p = 0.03) and less time in the heavy domain (p = 0.03) than the bottom 5 athletes (Table 3). Racing HR was lower on days 5 (115 ± 9, p = 0.01) and 6 (117 ± 21, p = 0.04) compared to day 1 (132 ± 12) across athletes (Table 4).

Average racing interstitial fluid glucose was 6.6 ± 0.5 mmol·L⁻¹. There was no statistical difference in active interstitial glucose concentrations between the top and bottom athletes (p = 0.2; Table 3) or between days across athletes (Table 4).

Environmental Conditions and Core Temperature

While variable throughout the day, average conditions across the 6-day race were 18.5 ± 1.7℃ , 44.3 ± 9.9% humidity, wind speeds of 3.7 ± 0.9 kph, and a UV index of 1.4 ± 0.5. Changes in ambient temperature and humidity across each day are illustrated in Fig. 3. Average precipitation was 2.8 ± 1.5 mm per day, with 4 of the 6 days receiving 2–4.5 mm of rain. Active core temperature was 37.6 ± 0.2℃ ranging from 37.4 to 37.8℃ and was weakly correlated with ambient temperature (r = 0.12, p < 0.001), and exercise intensity as %HRmax (r = 0.26, p < 0.001). Average core temperature during active movement was highest in the afternoon (37.7 ± 0.3℃), followed by overnight (37.6 ± 0.2℃), and was lowest in the morning (37.5 ± 0.2 ℃; all p < 0.001). Peak core temperatures were all < 39℃ during the race (average 38.5 ± 0.3℃, range 38.0 to 38.8℃). Athlete #1 had an average core temperature of 37.8℃, which was the same as Athletes #2 and #9 (Table 3). There was no difference in racing core temperatures between the top and bottom athletes (p = 0.8) or between days across athletes (Table 4).

In-Race Biomechanics

Step counts reflect race-related activity, as every athlete was transported (via golf cart) any time they were off-course. Total step counts over 6 days ranged from 196,625 to 898,618 steps (Table 3; Fig. 2). Running step counts over the event ranged from 4104 to 860,439 running steps. Athlete #1 averaged 149,770 total steps per day, of which 95.8% (143,407 steps) were classified as running steps, while the athlete who completed the second farthest distance averaged 114,879 total steps per day (93.9%/107,887 running steps). The average total daily steps for Athletes #3–5 was 87,667, of which 55.1% (46,342) were classified as running steps. Athlete #1 dropped from an average of 97.9% running steps over the first 3 days to 91.8% for the last 3 days, whereas Athlete #2 dropped from 98.1% to 87.7%. Athletes #3–5 averaged 56.4% running steps over the first 3 days and dropped to 41.3% over the last 3 days. Overall, Athlete #1 maintained a minimum of 96,771 running steps per day, while Athletes #2–5 had minimums of 17,359, 15,430, 0, and 16,974 running steps per day, respectively. Athlete #4 did not have any running steps on day 3 due to a transient injury.

When analyzed without adjusting for velocity, Athlete #1 displayed consistently longer stance times, a lower step rate, lower peak vertical ground reaction forces (GRFs), and higher vertical impulses compared to the other top finishers, particularly from days 3–6 (Table 3). These differences diminished when velocity was included as a covariate, highlighting that the slower average velocity maintained by Athlete #1 was a defining feature of her strategy and biomechanics. Comparisons between the top five and bottom five finishers revealed significant differences in stance and flight times across the event, with the top finishers maintaining longer stance times and shorter flight times, indicative of a more terrestrial gait pattern. Across the 6-day race, there were no significant effects of day on group mean biomechanical variables, regardless of whether running velocity was included as a covariate (Table 4).

Total Time Spent Sleeping/Recovery

Sleep data were unavailable for one athlete (Athlete #5), and incomplete for three other athletes over 1–3 days (Fig. 2). Of the complete data (n = 6), average hours slept per day was 5.2 ± 1.9 h, with no difference between days across athletes (Table 4). Athlete #1 slept the least at 1.5 ± 1.1 h per day, while the next top athletes slept 5.6 ± 0.1 h per day compared to the bottom finishers at 6.2 ± 0.5 h per day (p = 0.2 for top vs. bottom finishers; Table 3; Fig. 2). Time spent in recovery, which includes sleep, ranged from 6.9 to 20.2 h per day across athletes (mean 12.8 ± 3.5 h per day).

Dietary Intake

Mean energy intake for each athlete ranged from 2,915 to 6,856 kcal·day⁻¹ between athletes (mean: 4188 ± 1109 kcal·day⁻¹) across the 6-day ultramarathon. Among individual days and athletes, the lowest energy intake recorded was 2362 kcal·day⁻¹ (Athlete #10) and the highest energy intake was 7,904 kcal·day⁻¹ (Athlete #1). Athlete #1’s daily EI was 6,856 ± 786 kcal·day⁻¹ compared to Athletes #2–5: 4,443 ± 919 kcal·  day⁻¹ and Athletes #6–10: 3442 ± 624 kcal·day⁻¹ (p = 0.04 for top vs. bottom finishers).

Energy Expenditure

TDEE ranged between 3727 to 11309 kcal·day⁻¹ across the 6-day ultramarathon (Fig. 1). Athlete #1 had a race TDEE of 11,309 kcal·day⁻¹ compared to Athletes #2–5: 6902 ± 1,741 kcal·day⁻¹ (range: 4,797 to 8,832 kcal·day⁻¹) and Athletes #6–10: 5613 ± 1271 kcal·day⁻¹ (range: 3727–7152 kcal·day⁻¹). There was no difference in race TDEE between the top and bottom athletes (p = 0.1).

Muscle Damage andInflammatory Markers

Immediately post-race,  CK was 15-fold higher (Pre: 101 ± 57 vs. Post: 1,160 ± 1148 U·L⁻¹; p = 0.01), CRP was 52-fold higher (Pre: 0.6 ± 0.6 vs. Post: 11.9 ± 11.9 mg·L⁻¹; p = 0.01), and IL-6, IL-16, and IL-1Ra were all twofold higher (p < 0.001). Hematocrit (Hct) was reduced by approximately 7% post-race (Pre: 41.9 ± 4.3 vs. Post: 34.8 ± 3.1%; p = 0.001). Athlete #1’s post-race CK was 4,039 U·L⁻¹ which was 4.5-fold higher than Athletes #2–5 (1209 ± 933 U·L⁻¹) and 8.8-fold higher than Athletes #6–10 (545 ± 312 U·L⁻¹) (p = 0.1 for top vs. bottom finishers; Fig. 4A). Athlete #1’s post-race CRP was 23.7 mg·L⁻¹ compared to Athletes #2–5 (16.8 ± 17.1, range: 1.1 to 32.8 mg·L⁻¹) and Athletes #6–10 (5.6 ± 4.9, range: 0.7 to 12.9 mg·L⁻¹) (p = 0.1 for top vs. bottom finishers; Fig. 4D). Daily distance (R² = 0.538, p < 0.001; Fig. 4B) and average pace (R² = 0.190, p = 0.01; Fig. 4C) were positively related to in-race CK values, but not CRP (Fig. 4E, F).

Fig. 4.

Individual in-race CK (A) and CRP (D) across 6 days of ultramarathon running. Relationships between daily distance ran (B, E) and average pace (C, F) vs. CK and CRP, respectively

Neuromuscular and Cognitive Fatigue

Prior to the race, the ranges of dorsiflexor MVC torque and VA were 25.0 to 46.6 N·m and 98.5–100%, respectively. Contrary to expectations, none of these values were reduced after the race (19.7–40.8 N·m and 91.9–100%, respectively; p ≥ 0.1). Athlete #1 had a 30% reduction of MVC torque, whereas the decrease was 3% for Athletes #2–5, and 9% for Athletes #6–10 (p = 0.971 for Athletes #1–5 vs. 6–10). VA was 1% lower for Athlete #1, unchanged for Athletes #2–5, and 2% lower for Athletes #7–10 (p = 0.252 for Athletes #1–5 vs. 7–10). None of the three tasks had outcomes that were affected by the race (p ≥ 0.05). For example, the go/no-go task revealed mean reaction times ranging from 243 to 317 ms vs. 254 to 288 ms pre- vs. post-race (a change of − 10% for athletes #2–5 and + 4% for athletes #6–10; p = 0.3 between athlete groups).

Psychosocial Experiences: Coping with Stressors and Pain

Themes were identified within the interview and field note data pertaining to how the women coped with stressors and pain leading up to and during the race. In (dis)connection through ultrarunning, athletes recounted how ultrarunning itself was a coping strategy they employed to manage (inter)personal and professional stressors in their everyday lives, and to push themselves to achieve their goals. Ultrarunning, and the FURTHER event, afforded them the opportunity to take on a new challenge while also connecting with others and community, yet also time in solitude to process stressors by engaging with nature, which fueled their persistence in training and throughout the race. Despite the use of ultrarunning as a coping mechanism, injury and pain leading up to and during the 6-day race were central stressors necessitating consistent attention and management. The athletes positioned pain as a physical and psychosocial demand to cope with in training through a task-oriented approach including appropriate pacing, race planning, and by drawing on support from their crews, all in an effort to optimize their individual performances. In this way, athletes continuously attempted to ‘solve the problem’ of pain. During the 6-day race as pain increased and became central as fatigue set in, attending to pain using this task-oriented approach no longer became adaptive. The women then moved away from a focus on their individual performance, and re-framed pain as a collective experience of suffering across all athletes in the race, which allowed them to draw on each other’s strengths and move through and with pain to complete the race together.

Discussion

The purpose of this investigation was to characterize the physiological and biomechanical demands and psychosocial experiences and coping strategies of a 6-day ultramarathon across a diverse cohort of female runners, with a focus on exploring potential determinants of performance in this event. All athletes ran their personal farthest distances over 6 days, and 9 out of 10 athletes pushed themselves to their physical and/or psychological limits. The zero-attrition rate is noteworthy as typically 10–20% of athletes get injured leading up to target running events [53]. While the exercise intensity sustained in the race was low, with most of the effort performed in the moderate intensity domain and core temperatures sustained in a normothermic range, the athletes underwent an extreme physiological challenge as evidenced by the exceptionally high energy expenditure (three of the highest female TDEE reported to date; Fig. 1), sleep deprivation, circulating inflammatory biomarkers, and reported physical pain. Contrary to expectations and despite elevated levels of metabolic markers of muscle damage, mean impairment of neuromuscular function was not apparent for the dorsiflexors the day after the race. From a biomechanical perspective, the results highlight substantial variability in running volume and step count profiles across participants during the 6-day ultramarathon, with increasing walking steps as the race progressed. Athletes coped with the physical pain of the event by shifting from focusing on personally dealing with the pain (e.g., support from their crews) to embracing it as a part of the collective experience (e.g., through cognitive reframing).

From a performance perspective, the athlete who completed the most distance and broke the 6-day female world record did not have many physiological characteristics that separated them from the other top finishers. This top athlete, however, had the highest EI, TDEE, and slept the least across the 6-day race. While Athlete #1 experienced a lesser decline in proportion of running steps, they had the greatest drop off in pace from the start to the end of the race, and the greatest indicators of physical breakdown. In general, the top finishers adopted a more grounded, durability-oriented gait, characterized by relatively longer stance phases and shorter flight times. In contrast, the lower finishers retained shorter contacts and longer flights, consistent with a lighter, more spring-like running style that appeared less sustainable over 6 days. Thus, contrary to shorter distance events, where fatigue resistance is a key indicator of performance [10], world-class 6-day ultramarathon performance may be dictated by the ability to persevere through extreme fatigue.

6-Day Compared to Other Ultramarathon Races

To date, relatively few research studies have systematically examined the physical and psychological demands of ultramarathon events exceeding 161 km or duration-limited events over 24 h [11], although existing work primarily in male athletes from races such as Tor des Géants [15, 54, 55] and UTMB [56] have contributed important insights. In FURTHER, the exercise intensity indicated by pace and HR (mean ~ 6.8 kph and 121 bpm) was lower than that reported for male and female participants in the 89 km Comrades Ultramarathon (~ 8.6 kph; 148 bpm) [57], and hilly trail races of 50 km (7.9 kph; 150 bpm) and 80 km (6.9 kph; 138 bpm) [58], but similar to that of 161–167 km events including the Western States Endurance Run (6 kph; no HR) [59], the Sulphur Springs Ultramarathon (6.3 kph; 119 bpm) [58], and the Ultra-Trail du Mont-Blanc (5.9 kph; 132 bpm) [60]. However, percent time spent in the heavy-to-severe intensity HR domains was greater in this 6-day race than in 161 km trail races (~ 14% vs. ~ 5%) [58, 61]. Overall, the athletes had a reduction in running velocity across the 6 days, which is consistent with the positive pacing demonstrated in 24 h [62], 100 mile [63], and 106 km events [60], whereby athletes slow down over the course of the event. The top athletes had greater reductions in running velocity from the first 2 days to the last 2 days compared to the bottom athletes, which is contrary to previous findings that more experienced athletes have more even pacing profiles [60, 62, 63]. However, some athletes (Athletes #4, #8, #10) did have a U-shaped or negative pacing profile that reflected easier days in the middle of the race. As the athletes ran on a flat course with a relatively low thermoregulatory challenge and access to rest, nutrition, hydration, and cooling every lap, this likely allowed them to race at similar, or even harder, intensities than those experienced during shorter races with greater environmental and nutrition/hydration challenges.

The athletes in this race experienced elevations in key markers of muscle damage and systemic inflammation immediately post-race, but the inflammatory response was less than that demonstrated in shorter events, such as the 130 km mountain ultramarathon in Vienna [64], and 50–160 km hilly trail runs [61]. Similarly, the elevation in CK as a marker of muscle damage post-race, while still a 15-fold increase, was lower than other shorter ultramarathons including Leadville 100 (mi) [65], and a 200 km run [66]. CK is augmented with longer distances and eccentrically biased downhill running [67]. There was minimal elevation change in FURTHER, but we did see that longer distance was related to greater creatine kinase concentrations, with a weaker association between running pace and CK concentration (Fig. 4). The lower than expected creatine kinase values may reflect the flat terrain, greater opportunity for clearance over 6 days [66], and the all-female cohort as females may incur less muscle damage and neuromuscular fatigue following ultramarathons compared to males [68, 69].

Importantly, muscular damage and fatigue-related impairments of muscle force have been proposed as the primary limiter to ultramarathon performance [10]. While we could not assess neuromuscular fatigue immediately following the race, 24 h later there was no mean reduction of MVC torque or VA for the dorsiflexors across participants. The dorsiflexor data were unexpected as previous studies, albeit in mostly male participants, report reductions of 24–41% for MVC torque and 6–29% for VA of the knee extensors or plantar flexors immediately following ultramarathon [70]. This impairment of strength and central fatigue for lower limb muscles was demonstrated to persist for > 48 h, with a ~ 10–14% reduction of MVC torque and ~ 2% reduction of VA two days after the 166-km Ultra-Trail du Mont-Blanc [71]. When tested ~ 30 min following the 330-km Tor des Geants, the magnitude of neuromuscular fatigue and muscle damage was smaller compared to shorter distance ultramarathons, which the authors attributed to lower mechanical stress from decreased racing intensity and more time walking [72]. For the present race, the relatively low intensity, the flat course, and possibly the sex of the participants [69] could explain the absence of lower limb impairments. In addition, the kinetics of in-race CK illustrate a resolution and ‘time adaptation’ of muscle damage in response to a high constant daily workload; therefore, the 24 h post-race neuromuscular fatigue tests may not reflect this response (Fig. 4; [73]). Finally, there was no evidence of cognitive fatigue 24 h after the race; however, this was not entirely unexpected as the athletes may have recovered over 24 h, and others have shown both impairments [74, 75] and improvements [76] to reaction time and cognitive function immediately following ultramarathon events ranging from 93.8 to 422.1 km.

Total energy expenditure in this event was exceptionally high compared to previously documented expenditures. In fact, the three highest TDEE values recorded in our sample represent the highest documented in females to date [77]. Ultramarathon events elicit high energetic demands, with multi-day ultramarathons presenting a particularly unique physiological challenge. Existing literature on multi-day ultramarathons is limited, with reported TDEE values ranging from approximately 6000 kcal/day in prolonged events spanning weeks or months [78, 79] to over 9000 kcal/day in relatively shorter running events lasting 2.5–4 days [80]. However, these data are primarily derived from male participants, and females typically exhibit lower TDEE due to lower body weight and fat-free mass [81]. Despite the caloric deficit, exercising glucose was generally maintained within a normoglycemic range (70–140 mg·dL⁻¹; Robertson et al., in-preparation), which is consistent with data from 156km and 24h trail races [82, 83]. Further investigation into the energetic limits of female ultramarathoners—and how this may relate to energy and carbohydrate intake, gastrointestinal tolerance, athletic performance, and health—is needed to better support female athletes participating in extreme endurance events.

Performance Determinants of 6-Day Female Ultramarathon

In regular distance running events (≤ 42.2 km), the primary determinants of performance include VO_2max, the ability to sustain a high percentage of VO_2max (e.g., critical velocity/second lactate threshold), running economy, and a recently proposed fourth metric: resilience/durability—or the ability to resist fatigue and maintain performance [84]. In long ultramarathon events, the first three primary determinants of performance may not be as important in predicting performance compared to regular distance events [85]. Investigations have shown that while VO_2max, running economy, and %VO_2max at the second threshold may predict 50–68 km ultramarathon performance, they generally do not predict 121–161 km performance [1, 9, 85]. Durability has not been studied extensively in performance literature but is very likely a key determinant of ultramarathon performance [10], and has been shown to distinguish top finishers from bottom finishers in a Backyard running event [74].

We chose to compare Athlete #1, who defined performance in this event by breaking the 6-day female world record and 11 other world records, to the next 4 top finishers to explore determinants of performance across a sample of competitive female ultramarathon runners. In accordance with previous findings [9, 10, 85–88], there were no clear differences between measures of running fitness (V̇O_{2 peak}, running economy), anthropometrics (total body mass, fat-free mass, body fat percentage), or age between the top runner and the next 4 athletes. Athlete #1 did have the highest %V̇O_{2 peak} at RCP compared to the others, although it was only 1% higher than Athlete #4. As the athletes did not necessarily push to volitional exhaustion in the treadmill test, these data may simply be reflective of these two athletes not reaching their true max. Training in the lead up to the event was unfortunately not documented; therefore, we cannot comment on whether Athlete #1 had a greater training volume than the other athletes. Furthermore, as HR data for Athlete #1 were lost, we cannot compare their HR-to-speed ratio as a marker of durability as has been used previously [74]. What is clear, however, is that Athlete #1 displayed high levels of neuromuscular fatigue and physical breakdown by the end of the event; together with the greatest time on course, the least sleep, but also the greatest energy intake. Athlete #1 had CK levels that were 4.5-fold higher than the next 4 finishers. The drop-off in pace from the first 2 days to the last 2 days of racing was 13% higher in Athlete #1 compared to the next top finishers, and Athlete #1 slept only an average of 1.5 h/day. In previous investigations of 24 h and 100-mile races, it has been shown that the fastest runners typically have the most even pacing strategies [62, 63]. Herein, the pacing strategy employed by Athlete #1 was to go out very fast to break the 72 h world record, and then hold on for the following 3 days. It is impossible to know if she would have completed more distance over 6 days with a more even pacing strategy, as she still was successful in breaking the 6-day world record at the time of the event. Athlete #1 also matched for the highest active core temperature and expended > 2477 to 6512 kcal per day more than the next 4 finishers, and consumed > 1726 to 3131 kcal per day more than the next 4 finishers. As such, it is evident that despite the benefits of physiological durability, in multi-day duration-limited events, such as the 6-day ultramarathon, the athlete who is able to push through the most fatigue, and take the fewest breaks, will likely be the top performer. This further suggests, as reflected by the findings from psychosocial observations and interviews, that psychological and social factors including adaptive coping and pain-management strategies, personal motivations, and (social) support resources including community and belonging, are likely the key determinants of performance in 6-day ultramarathon races.

From a biomechanical perspective, the athlete who completed the greatest distance (Athlete #1) demonstrated remarkable consistency by maintaining a high percentage of running steps relative to total steps (95.8%), though she did experience a noticeable decline during the final 3 days (from 97.9% to 91.8%). In contrast, athletes who covered less distance (Athletes #2–5) exhibited substantially lower proportions of running steps overall (64.8%), with a sharper decline in the second half of the event (from 66.8% to 52.9%), suggesting greater reliance on walking as fatigue accumulated [89]. The capacity to maintain a running gait, even with progressive neuromuscular fatigue and physical pain, could be a significant determinant of success [84]. In addition, the sharper decline in running step proportions observed among the other top finishers underscores how biomechanical adaptations, potentially driven by pain, injury avoidance strategies, or neuromuscular fatigue, may directly influence pacing strategies and overall race performance [90].

A defining feature of the top finishers was their ability to sustain a running gait for the majority of the race, albeit with gait mechanics that correspond with what Lussiana et al. have termed a “terrestrial running form,” characterized by a higher duty factor (longer stance time, shorter flight time) [91, 92]. Athlete #1 exemplified this terrestrial pattern, maintaining longer stance times, a lower step rate, lower vertical GRF peaks, and higher vertical impulses compared to the other top finishers, especially in the latter half of the event. These biomechanical features are often associated with slower running [93]—and indeed, did result in slower average running paces than Athletes #2–5 over the last 4 days. However, in the context of a 6-day ultramarathon, this biomechanical approach may have reduced per-step mechanical loading, potentially limiting neuromuscular fatigue and damage [94], and enabled Athlete #1 to accumulate an exceptionally high number of running steps. The observation of a terrestrial running style aligns with previous observations during ultra-endurance events, where athletes adopted a running form illustrated by decreased vertical GRF peaks and decreased flight times as the event progressed [95, 96]. By limiting the vertical oscillation of the centre of mass, this running form favours durability and forward progression [91]. Although a terrestrial running style may be less economical at higher running speeds, at slower endurance paces, it can reduce mechanical stress and distribute loading more effectively, supporting sustainable performance over prolonged durations [94]. By comparison, the bottom five athletes maintained shorter stance times and longer flight times, characteristic of a faster, but potentially less sustainable running style. Importantly, these differences emerged despite no group-level differences in treadmill-measured running economy, underscoring that performance in a 6-day ultramarathon is not determined by laboratory measures of efficiency or by preserving a classic spring-like running gait, but rather by employing mechanics that prioritize durability over speed. Athlete #1 represented the most extreme case of this approach, demonstrating that slower, more terrestrial biomechanics can support world-record performance when sustained over prolonged periods.

These observations suggest that high-level performance in a duration-limited event does not depend on preserving a traditional, spring-like running pattern, typically associated with better running performance in shorter distance-limited events, but rather on adopting mechanics that prioritize durability and load management. These biomechanical profiles aligned closely with the physiological outcomes. Longer stance times and elevated impulses may have reduced acute musculoskeletal strain, allowing for continuous running, but they also coincided with the highest energy expenditures, greatest creatine kinase elevations, and the least sleep among the cohort over the duration of the event. Thus, while the strategy enabled maximal distance, it also came with the most severe physiological stress.

Finally, Athlete #1 was clearly the most experienced ultramarathon runner in this cohort, with 29 years of running experience, a 100 km personal record time of 7:08, and their furthest race prior to this event was 434.5 km. By contrast, the next top finishers had 4–18 years of experience, 100 km personal records of 8:02 to 14:16, and furthest prior race distances of 161 to 326 km. As such, Athlete #1 had much more experience to draw upon, and likely a better ability to frame pain and pacing strategies, across this event.

Strengths and Limitations

In this field-based investigation, we aimed to maximally support world-class ultramarathon preparation and execution in an attempt to push the limits of female ultra-endurance performance while also performing best practice field-based research methods that were minimally intrusive to the athletes [21]. A key strength of this study was the ability to capture biomechanical and physiological data continuously across 6 days of racing, a context in which very few previous investigations have been performed. This allowed us to examine not just pre- vs. post-race changes, but the evolution of biomechanical and physiological responses in real time. Furthermore, detailed investigations of world-record performances are extremely rare in the field of sports science.

From an applied perspective, these results provide practical insights for female athletes, coaches, and sports scientists, particularly regarding pacing strategies and how exercise intensity, physiological strain, and fatigue evolve across multi-day ultra-endurance events. Continuous monitoring allows for a detailed examination of real-time changes in key variables, which may inform training design, in-race decision-making, and recovery strategies in female ultramarathon contexts.

The primary limitations to this report include the descriptive nature of the results due to the heterogenous cohort, which required a ‘case-study’ approach. The small sample size limits statistical power and generalizability of the results; therefore, we have interpreted results in terms of practical significance. Due to the small heterogenous cohort and 6-day race window, we were unable to draw conclusions regarding menstrual cycle status and the outcomes presented herein. There are missing data mostly due to technical limitations and logistical challenges relevant to real-world data collection (e.g., battery life, data transfer, and sensor malfunction), environmental conditions, loss of contact with the athlete, and/or athlete’s choosing to opt-out of certain measures. Accordingly, we have transparently reported total sample size for each outcome. While the reliability of the waist-mounted IMU used in this study was not directly assessed, previous work has demonstrated good-to-excellent short- and long-term reliability of acceleration-derived metrics from the same system when embedded in footwear [97]. Furthermore, a recent systematic review reported that IMUs placed at the sacrum/lumbar region provide valid and reliable estimates of key spatiotemporal variables during running, supporting the use of this placement for field-based gait analysis [98].

The highly selective nature of the cohort (i.e., pre-selected world class and recreational athletes participating in a record-focused event) introduces potential selection bias. In addition, the exceptional race setup—including attempts at multiple records, individualized pacing, fuelling, and sleep strategies, and substantial logistical support—may limit the external validity of the findings. The time spent in HR intensity domains was not adjusted for cardiovascular drift or fatigue across the event, as this will require in-depth future analysis. We are also unable to disclose individual psychosocial data as it would breach participant confidentiality.

In this race, Athlete #1 epitomized peak performance; however, it remains unclear whether they might have achieved an even better outcome with different pacing and sleep strategies had they not also been pursuing additional world records within the greater attempt, including the 72 h record. Finally, a secondary exploratory component of this paper was a comparison between the top athlete and the next four finishers. World-class performance was primarily assessed as the top performer vs. the next 4 finishers as they represented a cohort of Tier 3 & 4 athletes. Forthcoming investigations from this data set, including specific comparisons with energetics, dietary analysis, and biomechanics will explore this question further.

Conclusion

Taken together, this data set is reflective of diverse female physiology under extreme physical stress in a real-world setting. The physiology of FURTHER was such that the exercise intensity sustained was similar to that of single-day mountainous events, but extended over 6 days. Exercising energy expenditure was extremely high, and in the case of three athletes, broke the highest female energy expenditures recorded to date. The markers of muscular breakdown and neuromuscular fatigue were less than expected, and may reflect the lack of downhill running or less intensity from the slower running/walking paces. Biomechanically, the ability to sustain a higher proportion of running steps throughout the event, despite accumulating fatigue and physical breakdown, emerged as a notable feature distinguishing the top performer. Performance appeared to depend less on preserving a traditional spring-like running gait and more on adopting a durability-focused strategy. The top finishers maintained longer stance times and shorter flight times, indicative of a more terrestrial style that may have limited acute loading while enabling continuous running. While completely exploratory, the main determinant of performance that set the top athlete apart from the other athletes of similar performance calibre was the amount of physical breakdown they sustained, which suggests that the ability to withstand pain and endure suffering may be the primary determinant in this very long event.

Abbreviations

BV

Blood volume

BW

Body weight

CGM

Continuous glucose monitoring

CO

Carbon monoxide

CK

Creatine kinase

CRP

C-reactive protein

DI

Dietary intake

DLW

Doubly-labelled water

DXA

Dual-energy X-ray absorptiometry

EDTA

Ethylenediaminetetraacetic acid

EI

Energy intake

FDR

False discovery rate

GET

Gas exchange threshold

GPS

Global positioning system

GRF

Ground reaction force

Hb

Haemoglobin

Hct

Hematocrit

Hbmass

Haemoglobin mass

HR

Heart rate

IMU

Inertial measurement units

LEAF-Q

Low energy availability in females questionnaire

MVC

Maximal voluntary contraction

OA-ICOS

Off-axis integrated cavity output spectroscopy

PR

Personal record

PV

Plasma volume

QC

Quality control

RCP

Respiratory compensation point

RCV

Red cell volume

SST

Serum separating tube

TDEE

Total daily energy expenditure

tHBmass

Total haemoglobin mass

VA

Voluntary activation

V̇O_{2 peak}

Peak oxygen consumption

V̇O₂ max

Maximal oxygen consumption

Author Contributions

T.S., R.J.G., J.F.B., conceived original study design. H.G.C., T.S., R.J.G., A.H., S.A.P., C.N., C.J.M., E.V.B., S.D.H., S.A.C., C.D.B., S.K.T.G., A.A.R., J.F.B., and A.M.C., performed data collection. All authors contributed to data analysis. H.G.C., T.S., R.J.G., A.H., S.A.P., C.N., C.J.M., E.V.B., J.F.B., and A.M.C. contributed to original draft preparation. All authors reviewed and edited the manuscript.

Funding

lululemon athletica inc. Mitacs Accelerate (IT39862), NSERC Alliance (#F24-01331), NSERC Alliance (ALLRP 592366-23).

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics Approval and Consent to Participate

This study was approved by the University of British Columbia clinical ethics review board (#H23-03110 and #H23-02179) in accordance with the Declaration of Helsinki, except for registration in a database.

Consent for Publication

Not applicable.

Competing Interests

lululemon athletica inc. was the majority funder of this event and associated research project, including support for travel, accommodation and incidentals, and supporting several trainees through Mitacs and NSERC programs (see below). There was no direct individual honorarium/compensation made beyond this, and further declared conflicts of interest.