Differences in pacing during cycling and running in ultra-triathlons - The example of 'Swissultra'.

OBJECTIVE: The knowledge of the most predictive split discipline and the pacing during a triathlon race is of utmost importance for planning an ultra-triathlon race. This study aimed at investigating the pacing during cycling and running splits in three different multi-stage ultra-triathlon race formats (i.e., Quintuple, Deca, and Double Deca Iron ultra-triathlon with 5x, 10x and 20x the daily distance of a full Ironman-distance triathlon). SUBJECTS AND METHODS: A total of 48 male ultra-triathletes competing in Quintuple (n=14), Deca (n=25), and Double Deca (n=9) Iron ultra-triathlon "swissultra" between 2016 and 2019 in Switzerland were analyzed. For each race day, we calculated the total performance (sum of all laps time), average individual performance (average of all laps time within a race day) and pacing variation (coefficient of variation of race laps time) for cycling and running. Discipline (cycling and running) and race distance (Quintuple, Deca, and Double Deca Iron ultra-triathlon) were used as independent parameters. The primary outcome variables were the time performance (daily and total) and the pacing variation. We applied two general linear models (GLM): the first model was a one-way ANOVA comparing total and daily performance by race distance, and the second model was a two-way ANOVA (race distance´ discipline) using pacing variation (average pace oscillation) as a dependent variable. RESULTS: The first GLM identified a significant race distance effect for total performance in both cycling (F = 375.6; p < 0.001; ηp2 = 0.943) and running (F = 267.8; p < 0.001; ηp2 = 0.922) with Double Deca Iron ultra-triathlon being the fastest and Quintuple Iron ultra-triathlon the slowest. The GLM for daily average performance showed no significant effect of race distance on cycling performance (F = 0.171; p = 0.843; ηp2 = 0.008), but on running performance (F = 6.408; p = 0.004; ηp2 = 0.222). The two-way GLM comparing pacing variation showed a significant effect for race distance (F = 11.81; p<0.001; ηp2 = 0.344) with Deca presenting larger pace variation than Quintuple and Double Deca Iron ultra-triathlon in both cycling and running, but not for discipline (F = 0.067; p=0.797; ηp2 = 0.001), nor for interaction (F = 1.469; p=0.241; ηp2 = 0.061). CONCLUSIONS: Athletes achieved a stable cycling performance independent of the length of the race, and the cycling split had an influence on the subsequent running split depending upon the length of the race.