Ethical approval

This study was approved by the Institutional Review Board of Kanton St. Gallen, Switzerland, with a waiver of the requirement for informed consent of the participants as the study involved the analysis of publicly available data (EKSG 01/06/2010). The study was conducted in accordance with recognized ethical standards according to the Declaration of Helsinki adopted in 1964 and revised in 2013.

Data set and data preparation

A dataset with 944,815 records, including professional and age group triathletes’ records, was downloaded from the official IRONMAN^® website (www.ironman.com) using a Python script (www.python.org). Each record included the athletes’ full name, sex, age, country of origin, event status, event location and year, the times for swimming, running, cycling, and transition times (represented by transition 1—swimming for cycling, and transition 2—cycling for running) and the full race time (all in seconds). The IRONMAN^® website describes a new the possibility that any athlete can compete as a transgender athlete. However, these athletes have to follow special rules (www.ironman.com/news_article/show/1258804). Furthermore, all data regarding name, sex, age, and country of origin are self-reported by the athletes when entering the race.

After isolating the age group records and pre-processing the dataset, a total of 687,696 valid finishers’ records, of which 553,608 were from men and 134,088 from women IRONMAN^® age group triathletes, from races between 2002 and 2022 were available for analysis. The athletes’ records were sorted in 5-year age intervals (except the first and last groups) conforming the age groups 18–24, 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70–74, 75+. Due to the low numbers of finishers in age groups 75–79, 80–84, and 85–89 years, athletes from these age groups were pooled into an age group of 75+ years. Exclusion criteria were (i) athletes who did not start or finish, (ii) disqualified athletes, (iii) records with missing split time, (iv) records with inconsistent times (i.e., impossible split times or final times smaller than split times, etc.) and (v) records without essential information (i.e., sex, age group, etc.). Location data information was added, including average temperatures for air and water and the type of swim (bay, river, ocean, lake, reservoir), bike (flat, rolling, hilly), and run (flat, rolling, hilly) course. This information was merged with the race data, in order to perform analyses of performance with the race course characteristics as covariate.

Statistical analysis

Trends over the year were calculated and plotted, including participation, men-to-women ratio, and average/ best performance. Histograms of the split and overall race times by sex were plotted, and density distribution curves were overlapped, showing an approximately normal distribution in all cases. Violin plots were used to compare the split and overall race times between age groups and sex, and boxplot charts were used to investigate the performance differences between sex by course types. Detailed data tables were also calculated and included in the study for additional detail, including a large summary table of the different IRONMAN^® event locations analyzed through the 20-year period. Statistical significance of the differences between groups was tested through ANOVA two-way and Tukey HSD post-hoc tests. The percent of change between men’s and women’s average times in each age group was also calculated and plotted for each split discipline and overall race times. Descriptive statistics are presented using mean, standard deviation, frequencies, minimum and maximum values, and percentages. The main variables object of study, the full or partial race times, originally recorded in seconds, is converted to hours and further processed with the format HH:MM:SS (hours:minutes:seconds) for presentation in tables. All analysis was done in a Colab Notebook (https://colab.research.google.com/) with Python (www.python.org/) and associated libraries.

Men were faster than women

A first important finding was that men were faster than women for all split disciplines and overall races times and for all age groups. Since women were not able to outperform men in the older age groups, we could not definitely confirm our hypotheses that women would close the gap to men in swimming and cycling with increasing age. However, the U-shape indicates a small closing of the gap between women and men at both the low and high ends of the age axis.

A potential explanation for this finding could be the fact that these IRONMAN^® age group triathletes were not specialists in swimming and cycling since they must compete in all three disciplines [35]. A further explanation could be that IRONMAN^® age group triathletes do not compete in the very old ages, such as master swimmers in pool swimming [15, 36] or master cyclists in ultra-cycling [17]. For pool swimming, it has been shown that women swimmers older than 75 years could achieve a similar performance to men. In master swimmers competing in the FINA World Championships in pool swimming, male age group backstroke swimmers were faster than female swimmers in age groups 25–29 to 80–84 years, but not in age groups 85–89 to 95–99 years [32]. In age group butterfly swimmers, women were not slower compared to men in the master group 90–94 years [15]. In age group breaststroke swimmers, men were not faster than women for age groups 90–94 to 95–99 years [3]. In age group freestyle swimming, women were slower than men in age groups 25–29 to 75–79 years, but not in age groups 80–84 to 85–89 years. In individual medley, men were faster than women from 25–29 to 80–84 years, but not from 85–89 to 90–94 years [36]. And in the FINA World Masters Championships in 3000 m open-water swimming, women were slower in age groups 25–29 to 70–74 years but not in age groups 75–79 and 85–89 years, where race times were similar for both women and men [16].

A subsequent explanation could be the number of women and men finishers in the older age groups. We had only one man in the age group 85–89 years during these two decades, but no women. No athletes older than the age group 85–89 years finished an IRONMAN^® triathlon in the last decades. This sex difference in athletic performance is due to physical (e.g., body size, body composition, length of limbs, running biomechanics), physiological (e.g., fat mass, muscle mass, muscle tissue characteristics, muscle strength, neuromuscular fatigue, aerobic capacity, oxygen uptake, hormones), technical, thermoregulation, sociocultural, sport-specific, and psychological factors [13, 37, 38]. In ultra endurance swimming, the gap is now less than 5% [8]. Although men have a larger body size with more skeletal muscle mass, a lower percentage of body fat, and a greater maximal delivery of anaerobic and aerobic energy, women have a better capacity to metabolize fat, demonstrate better hydrodynamics, and adopt a more even pacing, which may be advantageous, in particular during long-lasting swimming competitions [8].

Sex differences in split disciplines

The sex difference in cycling was more significant compared to running, which was more significant compared to swimming. This finding is in contrast to existing reports that describe a less significant sex difference in swimming compared to cycling and running in both the IRONMAN^® distance triathlon [23] and the Olympic distance triathlon [24]. A potential explanation could be that our sample consisted of nearly 700’000 race records.

When we consider the single disciplines not performed within a multi-discipline event such as a triathlon, the sex differences were less significant in swimming compared to running [23] and cycling [39]. Similarly, when draft-legal ultra-distance events such as swimming and cycling were compared, the sex difference of the fastest athletes was smaller in swimming than in cycling [40]. In triathlon, the sex difference in split disciplines seemed to depend upon the discipline [41], the age of the athlete [27] and the length of a race [41]. Considering age, the sex difference increased after the age of 35 years in an Olympic distance triathlon [41]. There seem to be differences between Olympic and IRONMAN^® distance triathlons where in an IRONMAN^® triathlon, the sex difference increased after the age of 55–60 years [23, 27]. Future studies might investigate the physiological, anthropometric, and biomechanical differences between these race distances.

U-shaped sex difference across age groups

The sex difference was lowest in the age group 18–24 years for all split disciplines and increased in a U-shaped manner until age group 70–74 years. We found that from age group 35–39 years until 60–64 years, the sex differences were nearly identical in swimming and running. The green (run) and light blue (swim) lines overlap between 35 and 55 years completely and hardly vary from 0.5% to 60 years.

One might assume that the U-shaped curve in the sex difference might be due to the distribution of IRONMAN^® finishers by age group since athletes in the middle age groups represented most of the finishers and athletes in the younger and older age groups were under-represented. However, the men-to-women ratio increased across age groups indicating that men were overrepresented with increasing age compared to women. Therefore, the sex difference might be due to a selection bias from a lower number of women finishers in the older age groups compared to men finishers. It has been reported that the sex difference in endurance performance regarding age depends upon the participation [22, 41]. A U-shaped sex difference has also been found for age group triathletes competing in Sprint, Olympic, and IRONMAN^® 70.3 World Championships. The sex differences for overall race time were greatest in the youngest and older age groups for all three distances [42].

We found differences for the split disciplines after age group 70–74 years, where the sex difference decreased in swimming and running but increased in cycling with increasing age. After the age of 80 years, only 20 men and 1 woman successfully finished an IRONMAN^® triathlon. This might have influenced the sex difference in swimming for the age group 75+ years. Regarding the age of 70 years, we also have to consider the age-related performance decline. This performance decline starts at ~30–35 years [43] and remains linear until the age of ~70 years [43], but decreases exponentially after the age of ~70 years [44]. Therefore, from 70–74 years to 75–79 years, this age-related performance decline might explain the differences in the split disciplines. Regarding the age of ~35 years and the widening of the sex difference in performance, we need to consider the maternal age of women and the influence of childbirth occurring often at this age [45]. We also need to consider the impact of 9+ months of limited to no training on performance recover and return to baseline fitness levels following childbirth.

Our findings that the sex difference was lowest in age group 18–24 years might be explained by the low number of finishers and the low men-to-women ratio. However, a study investigating IRONMAN® 70.3 age group triathletes showed that the sex difference in performance was also U-shaped in swimming and running, with an increase after the age of 18–24 years in swimming and after 40–44 years in running. In contrast, the sex difference decreased continuously with increasing age for cycling [28]. Future studies might investigate the potential physiological, anthropometric, and biomechanical differences for this age group compared to older age groups.

Sex differences in age groups

For age group 75+ years, we found differences between the split disciplines, where the sex difference decreased in swimming and cycling but increased in running. When we consider the age-related performance decline, endurance performance depends upon age [39] and declines with increasing age [39]. For Olympic distance triathletes, the age-related declines in performance were significantly less pronounced for men compared to women for swimming (>50 years), for cycling (>40 years) and for overall race time (>40 years) [26]. A triathletes’ performance decreases in a curvilinear manner with advancing age [2] where the age-related performance decline seems specific to both the split discipline [2] and the race distance [2], where the age-related performance decline differs between race distances [24].

For single disciplines not combined within in multi-disciplines event, we know that women were not able to reduce the gap to men in several different sports disciplines [46]. In recent years, however, women have been able to reduce the gap with men in specific sports disciplines. This has been especially visible for long-distance cycling [47] and swimming of different strokes and distances [15]. In long-distance swimming, women can achieve a similar [16] or even better performance than men [38]. It has been shown that women can beat men in swimming such as long-distance open-water swimming [38].

With increasing age, women seemed to be able to reduce the gap in endurance performance to men [42, 43, 46]. Investigations in swimming, cycling, and running [22, 24, 25] showed that elderly women were able to reduce the gap to elderly men. Specifically, in distance-limited ultra-cycling races covering 100 miles, 200 miles, 400 miles, and 500 miles, the sex difference in cycling speed decreased with increasing age [43]. Also, in time-limited ultra-cycling races of 6 hours, 12 hours and 24 hours in duration, the sex differences in cycling speed decreased between men and women with increasing age [11]. In 50-mile and 100-miles running races, the sex difference decreased with higher ages and was smaller in 100-mile (4.41%) than in 50-mile races (9.13%) [20].

The decrease in the sex difference is most probably because more women compete at older ages [24] and these women are more competitive than men [48]. However, this participation depends upon discipline and race distance. Furthermore, elderly women after the age of 70 years seemed to better preserve muscle quality than men of the same age. It was also assumed that women runners have a greater fatigue resistance than equally trained men. A hormonal influence with advanced age should also be considered [49]. Sex hormones seem to influence differences in the prevalence of diseases, in the magnitude of aging, and in the longevity between men and women [50]. Estrogens have significant impacts on the central nervous system and after menopause, a decrease in estrogens might impact cognitive function [51]. However, towards the end of life, women are frailer and have worse health, while men still perform better [52]. These hormonal changes that occur post-menopausal in women are due to a decrease in estrogen [53]. Furthermore, sex differences do exist regarding musculoskeletal injuries and diseases between women and men at different ages [54].

The sex difference in performance with increasing age seems to depend upon the participation rates to the competitions under analysis. The sex difference in endurance running with increasing age has been described as being dependent upon the participation of women [19]. In ultra-marathon running, the sex difference was more significant when fewer women than men competed [19]. In long-distance running such as marathon [52] and ultra-marathon running [19], the lower participation of women compared to men has an influence on the sex difference in performance.

A further aspect is that we collected data from two decades and that women might have reduced the gap to men over this time. We have calculated and plotted the percent change in overall race time over the years. The finding suggests there may be a slight increase in the performance gap through the years of observation. While we analyzed data from 2002 to 2022, a study investigating race data from IRONMAN^® Hawaii from 1983 to 2012 showed that the sex difference for overall race time decreased over years [55]. For the split disciplines, the sex difference remained unchanged for swimming and cycling but decreased for running. Lepers stated in his 2019 review that nowadays, elite female IRONMAN® triathletes are able to reduce the gap with their male counterparts to less than 10% of total performance, thanks to improvement in their marathon running performance [24].

It has been shown that women could reduce the sex difference over years in swimming [25] and in cycling races [11]. The sex difference in running as a weight-bearing discipline seemed to differ compared to cycling and swimming as non-weight-bearing disciplines. The sex gap in running performance is also due to the performance level. The sex difference was least significant between women and men world records in running but increased in lower-ranked athletes [14]. We also need to consider that there exist sex inequalities [56] and training and coaching differences for female athletes on their way.

The aspect of distance

The reduction of the sex gap seemed to depend upon the length of an endurance performance. In running, the sex difference increased with increasing running distance [17]. In running races from 60 m to 10,000 m, the sex difference increased over distances [57]. For longer distances such as a half-marathon and a marathon, the sex difference was less significant for the marathon distance compared to the half-marathon [58]. In half-marathon and marathon running, the sex difference in performance was less significant in the longer race distances and the older age groups [58]. When 12- and 24-hour running races were compared, the 24-hour race showed a lower sex difference in performance [33]. This might also be due to participation differences. In ultra-cycling races from 6 hours to 24 hours, the sex differences in cycling speed decreased with increasing duration of ultra-cycling races [17]. In races from 100 miles to 500 miles, no sex difference was identified for 400 miles and 500 miles [33]. When trail runners of the same performance level were investigated, the gap between women and men became reduced with increasing running distance, demonstrating that endurance is greater in women compared to men [59]. In triathlon races, the sex difference increased from the IRONMAN^® distance from IRONMAN^® Hawaii to the double distance of the Double Iron ultra-triathlon [60] and the Deca Iron ultra-triathlon [48].

The aspect of participation and performance level

The men-to-women ratio in long-distance events seems important for the sex difference. In running races from 5 km to ultra-marathon covering a data set from more than 1,100,000 running race records, women were more prevalent in shorter races (i.e., 5, 10 km, half-marathon) and outnumbered men in 5 km races. With increasing race distance, the men-to-women ratio declined, and the sex difference decreased until 70 years of age, after which it varied depending on the race distance. Elderly female ultra-marathoners (i.e., 75 years and older) displayed a performance difference of <4% compared to male ultra-marathoners, most likely attributed to the presence of highly selected outstanding female performers [21]. The reduction of the sex difference with increasing age is most likely due to the lower overall participation [20] and the higher participation of elderly women [20] leading to a selection of the best female ultra-endurance athletes in the higher ages [21]. When female and male ultra-marathoners with comparable numbers in 50-mile and 100-mile ultra-marathons were compared, the sex difference was reduced to 1%-3% [61]. We also need to consider the aspect of performance level. In running events from 100 m to the marathon, the sex difference varies with the performance level. The difference in absolute running performance between men and women is lowest for world record/world lead performances and increases in lower-ranked elite athletes [14].

Limitations

Some limitations must be acknowledged regarding the variation of race characteristics across different locations (e.g., environmental conditions) and the advancement of technology during the last two decades. The data stems from different race locations where the distances of the split disciplines might not have been exactly measured. Furthermore, during the two decades, the technologies in cycling (i.e., time trial bikes) might have advanced considerably. The influence of environmental conditions and changes in altitudes were not considered. Also, caution is needed when interpreting the findings on the old age groups, where the number of women (12 in the age group 75–79 years and one in the 80–84 years age group) would not allow safe conclusions. We have to acknowledge the limitation of the men-to-women ratio for a comparative analysis. With a men-to-women ration between 3 and 4.5 we do not believe however the male over-representation to be excessive, but just a reflection of the competition demographics. A further limitation is that post-partum information was not collected nor accounted for in data analysis. And a last limitation was that we have no data about transgender athletes. Transgender athletes are considered in this study as the sex identify with at the time of the racing. On the other hand, this cross-sectional data analysis uses the most extensive data set to date regarding the analysis of the sex difference in IRONMAN^® age group triathletes. Another strength of the present study was that it used the model of triathlon to examine sex differences in three basic modes of human locomotion, and this methodological approach was better than separate swimming, cycling, and running races of different athletes.