Age Differences in Pacing in Endurance Running: Comparison between Marathon and Half-Marathon Men and Women

Background and Objective: The increased popularity of marathons and half-marathons has led to a significant increase in the number of master runners worldwide. Since the age-related decrease in performance is dependent on race duration, pacing in long distance running might also vary by race distance in both men and women. Therefore, the main aim of this study was to assess pacing differences between marathon and half-marathon runners with regard to the runners’ age group, and independently for men and women. Materials and Methods: In total, 17,465 participants in the Vienna City marathon in 2017 were considered for this study (marathon, N = 6081; half-marathon, N = 11,384). Pacing was expressed as two variables (i.e., pace range and end spurt). Results: All runners showed positive pacing strategies (i.e., a fast start with gradual decrease of speed). However, marathon runners showed greater variability in pacing than half-marathon runners. Furthermore, women showed no differences in pace variability in regard to the age group, whereas men younger than 30 years of age, as well as older men (over the age of 60), showed a greater variability in pace than other age groups. Finally, younger half-marathon men and women showed the fastest end spurt compared to older age groups and marathon runners. Conclusions: The presented findings could help sports and medicine practitioners to create age specific training plans and pacing strategies. This approach could help long distance runners to improve their physical fitness, achieve better race times, reduce the potential risk of musculoskeletal injuries and increase the overall pleasure of long distance running.


Introduction
Pacing in long distance sport events, such as cycling, running, cross-country skiing or triathlon races, can be defined as the efficient distribution of energetic reserves, power output, and speed through the entire race, without a significant slowdown [1]. For successful long distance pacing, it is necessary to select a suitable pacing strategy. This can be defined as a self-selected strategy that long distance athletes adopt from the beginning of a race [2]. Several pacing strategies have been previously observed by researchers [3,4]. These can be broadly categorized as: negative pacing (slow start with gradual increase of speed), positive pacing (fast start with gradual decrease of speed), even pacing (without significant speed changes) and variable pacing (with significant speed changes). Selecting an optimal pacing strategy can be a crucial aspect in the successful completion of a long distance event [5,6]. Additionally, the best pacing strategies can decrease the potential risk of musculoskeletal injuries [4], participants as the study involved the analysis of publicly available data. The study was conducted in accordance with recognized ethical standards according to the Declaration of Helsinki adopted in 1964 and revised in 2013.
For the purpose of this study, we included official results and split times from the 2017 Vienna City Marathon [24], i.e., the initial sample. Participants who did not finish any of the races, or did not have record of any of the split times were excluded from the initial sample. Moreover, participants who did not provide information on their age were also excluded from the initial sample. In total, 17,465 participants were considered for this study (marathon, N = 6081; half-marathon, N = 11,384).
Further information regarding the 2017 Vienna City Marathon is as follows: • Both the marathon and the half-marathon were held on the same day, on an officially certified and fairly flat track (the elevation difference was only 50 m; ranging from 154 to 210 m). For comparison, the Berlin Marathon, considered to be "the fastest marathon" has an elevation difference of 21 m [19].

•
During the race day, the weather was cloudy, with temperatures ranging from 7.8 • C at 9 am to 11.8 • C at 2 pm, without excess humidity or strong wind [24].

•
No additional information on humidity grade or wind speed was available on the official race website.

•
The half marathon race was entirely contained within the marathon race.

Data Analysis
In the first step of data analysis, we calculated the average speed for the entire race for each participant in both the half-marathon and the marathon. Additionally, we calculated the average running speed in five race sections, for both the marathon and the half-marathon [6,25]. The race sections were divided as follows: Section 1 included the average running speed from start to the 10th km of the marathon race and from start to the 5th km of the half-marathon race. Both distances correspond to the first 23.7% of the marathon and half-marathon races.
Section 2 included the average running speed from the 10th km to the 20th km of the marathon race and from the 5th km to the 10th km of the half-marathon race. These distances represent a section of 23.7-47.4% of the marathon and half-marathon races. Section 3 included the average running speed from the 20th km to the 30th km of the marathon race and from the 10th km to the 15th km of the half-marathon race. Both distances represent a section of 47.4-71.1% of the marathon and half-marathon races. Section 4 included the average running speed from the 30th km to the 40th km of the marathon race and from the 15th km to the 20th km of the half-marathon race. These distances represent a section of 71.1-94.8 % of the marathon and half-marathon races.
Section 5 (i.e., the end spurt), included the average running speed from the 40th kilometer, to the race finish (42.195 km) in the marathon as well as from the 20th kilometer to the race finish (21.0975 km) in the half-marathon. The end spurt represents a section of94.8% to the finish line of the marathon and half-marathon races.
Furthermore, considering the methodology of Breen et al. [26], we calculated the variable of interest, called the "pace range". Five race sections were subsequently expressed as a percentage faster or slower than the average section speed. The fastest section for each individual was then named the "positive range" (PR), while the slowest segment was named the "negative range" (NR). In addition, the absolute sum of the positive range and negative range was calculated and named the "pace range" (PaceR). This method allowed for normalized speed comparisons between all athletes as well as between the marathon and half-marathon.
Finally, to examine the final 2.195 km of the marathon and the final 1.0975 km of the half-marathon, running speed for this segment was expressed as a percentage faster or slower than the running speed during Section 4 (i.e., 30-40 km for the marathon and 15-20 km for the half-marathon). This variable aimed to examine the "end spurt" (ES) and was named as such.

Statistical Analysis
Prior to all statistical tests, descriptive statistics were calculated as the mean and standard deviation. Moreover, data distribution normality was verified by visual inspection of histograms and QQ plots [6].
To assess sex and age group distribution among the participants in the half-marathon and marathon, we used a chi-square test (χ 2 ). Specifically, we examined the association between participants' sex and age group separately for each race, as well as between their sex and the race they participated in. The magnitude of these associations was tested by Cramer's phi (ϕ), while the results were presented as a men-to-women ratio (MWR).
For all ANOVAs, the post-hoc Bonferroni test was performed. Effect size was presented via eta squared (ŋ 2 ), where the values of 0.01, 0.06 and above 0.14 were considered small, medium, and large, respectively [27]. Alpha level was set at p < 0.05. All statistical tests were performed using Microsoft Office Excel 2007 (Microsoft Corporation, Redmond, WA, USA) and SPSS 20 (IBM, Armonk, NY, USA).

Results
The number of men and women in each race (i.e., men-to-women ratio -MWR) and their age group are presented in Table 1. As expected, more men participated in the 2017 Vienna Marathon than women (Total MWR was 2.44). A sex x race association was shown, (χ 2 = 293.6, p < 0.01, ϕ = 0.13), where the men-to-women ratio was smaller in the half-marathon (1.99) in comparison to the marathon (3.77). Furthermore, sex x age group association was also shown in both the marathon (χ 2 = 72.4, p < 0.01, ϕ = 0.11) and half-marathon (χ 2 = 263.6, p < 0.01, ϕ = 0.15). In the marathon, the smallest MWR of 2.45 was observed in the 25-29 years age group, whereas the largest MWR of 8.35 was observed in the oldest age group. In the half-marathon, the smallest MWR of 1.14 was observed in the youngest age group, whereas the largest MWR of 3.96 was observed in the oldest age group. The average running speeds for four sections and ES are presented in Table 2. From the descriptive data in Table 2, we can observe a gradual decrease in average speed throughout the race segments for both sexes in both the marathon and half-marathon, among all age groups. Furthermore, we can observe that ES are typically faster than Section 4, which has often been noted in marathon related studies. When PaceR in men runners was assessed (Figure 1), the results showed significant main effects of race [F (17, When PaceR in men runners was assessed (Figure 1), the results showed significant main effects of race [F(17, 12366) = 849.8, ŋ 2 = 0.06, p < 0.01], age group [F(17, 12366) = 14.3, ŋ 2 = 0.01, p < 0.01] as well as race x age group interaction [F (17,12366) = 3.5, ŋ 2 < 0.01, p < 0.01]. Bonferroni post-hoc test results for each race showed higher PaceR in marathon runners in comparison to half-marathon runners for all age groups (p < 0.01). Regarding age groups, all marathon runners younger than 30 years of age showed significantly higher PaceR than all runners from 30 to 59 years of age (p < 0.01). Moreover, marathon runners from the 60+ years group had higher PaceR than runners from 35 to 49 years of age (p < 0.01). Finally, half-marathon runners younger than 24 had higher PaceR than all runners from 30 to 54 years of age (p< 0.01).
When PaceR in women runners was assessed (Figure 2), the results showed significant main effects of race (F(17,5063) = 79.3, ŋ 2 = 0.02, p< 0.01) and age group (F(17, 5063) = 3.5, ŋ 2 = 0.01, p = 0.01). A significant main effect of race x age group interaction was not obtained (F(17, 5063) = 0.62, ŋ 2 < 0.01, p = 0.76). Bonferroni post-hoc test results for each race showed higher PaceR in marathon runners in comparison to half-marathon runners for all age groups (p < 0.01). Regarding age groups, all marathon runners younger than 30 years of age showed significantly higher PaceR than all runners from 30 to 59 years of age (p < 0.01). Moreover, marathon runners from the 60+ years group had higher PaceR than runners from 35 to 49 years of age (p < 0.01). Finally, half-marathon runners younger than 24 had higher PaceR than all runners from 30 to 54 years of age (p < 0.01).
When PaceR in women runners was assessed (Figure 2), the results showed significant main effects of race (F (17,5063) = 79.3, ŋ 2 = 0.02, p < 0.01) and age group (F (17,5063) = 3.5, ŋ 2 = 0.01, p = 0.01). A significant main effect of race x age group interaction was not obtained (F ( Bonferroni post-hoc test results for each race showed higher PaceR in marathon runners in comparison to half-marathon runners for all age groups (p < 0.05). No significant differences were observed when the Bonferroni post-hoc test was performed for age group.
When the ES in men runners was assessed (Figure 3), the results showed significant main effects of race [F(17, 12366) = 25.5, ŋ 2 < 0.01, p < 0.01] and age group [F(17, 12366) = 13.6, ŋ 2 = 0.01, p < 0.01] Bonferroni post-hoc test results for each race showed higher PaceR in marathon runners in comparison to half-marathon runners for all age groups (p < 0.05). No significant differences were observed when the Bonferroni post-hoc test was performed for age group.

Figure 2.
Women's pace range (%) by age group for the marathon and half-marathon. Error bars represent 95% confidence intervals.
Bonferroni post-hoc test results for each race showed higher PaceR in marathon runners in comparison to half-marathon runners for all age groups (p < 0.05). No significant differences were observed when the Bonferroni post-hoc test was performed for age group.
When the ES in men runners was assessed (Figure 3), the results showed significant main effects of race [F(17, 12366) = 25.5, ŋ 2 < 0.01, p < 0.01] and age group [F(17, 12366) = 13.6, ŋ 2 = 0.01, p < 0.01] whereas a significant main effect of race x age group interaction [F(17, 12366) = 0.97 ŋ 2 < 0.01, p = 0.46] was not observed.  Bonferroni post-hoc test results for each race showed higher ES in half-marathon runners aged 25-49 years (p < 0.01). Moreover, half-marathon runners from 55-59 years of age showed higher ES than marathon runners of the same age (p < 0.05). Regarding Bonferroni post-hoc test results for age group, both marathon and half-marathon runners younger than 24 years showed significantly higher ES than all runners older than 30 years of age (p < 0.01). Moreover, half-marathon runners aged 24-29 years showed significantly higher ES than all runners older than 35 years of age (p < 0.01). Finally, half-marathon runners aged 30-34 years showed significantly higher ES than 45-49 year-old runners as well as the 60 + age group (all p < 0.01).
Regarding Bonferroni post-hoc test results for age group, age groups <24 and 25-29 years showed significantly higher ES than all runners older than 35 (p < 0.01) in the half-marathon. Furthermore, half-marathon runners from 30-34 years showed significantly higher ES than all runners older than 45 (p < 0.01). Finally, runners older than 60 years showed significantly lower ES (p < 0.05) than runners from 35 to 49 years of age. aged 24-29 years showed significantly higher ES than all runners older than 35 years of age (p< 0.01). Finally, half-marathon runners aged 30-34 years showed significantly higher ES than 45-49 year-old runners as well as the 60 + age group (all p < 0.01).

Discussion
The main aim of this study was to assess differences in pacing between marathon and half-marathon events in regards to the runners' age groups, separately for men and women. In both men and women, regardless of age group, marathon runners showed greater variability in pacing than half-marathon runners. Furthermore, women showed no differences in pace variability with regard to the age group, whereas men younger than 30 years of age, as well as older men (over the age of 60), showed greater variability in pace than other age groups. Finally, younger half-marathon men and women showed the fastest end spurts compared to the older age groups and marathon runners.
Although more men participate in long distance events than women, several recent studies have reported a significant increase in women participants over the previous decades [28,29]. In this

Discussion
The main aim of this study was to assess differences in pacing between marathon and half-marathon events in regards to the runners' age groups, separately for men and women. In both men and women, regardless of age group, marathon runners showed greater variability in pacing than half-marathon runners. Furthermore, women showed no differences in pace variability with regard to the age group, whereas men younger than 30 years of age, as well as older men (over the age of 60), showed greater variability in pace than other age groups. Finally, younger half-marathon men and women showed the fastest end spurts compared to the older age groups and marathon runners.
Although more men participate in long distance events than women, several recent studies have reported a significant increase in women participants over the previous decades [28,29]. In this study, total MWR was 2.44. Moreover, the MWR was smaller in younger age groups, particularly in the half-marathon, where the lowest MWR was 1.14. The rationale for this can be found in fewer opportunities for older women to participate in sports [30], whereas a greater number of younger women have started to engage in regular training and sport in recent years [28]. Furthermore, the higher MWR in marathons might be explained by social discrimination and stricter medical caution for women than for men [31]. In addition, motivational factors, such as aspirations to have fun and stay healthy could justify the increase in women participating in half-marathons in comparison to men [32,33].
As seen in Table 2, positive pacing strategies (i.e., running speed decreased through the race) were observed in both marathon and half-marathon events, in all men and women age groups, which is in line with similar studies [19,22]. Moreover, positive pacing strategies often involve end spurts in the final sections of marathons and half-marathons [6,34], which was observed in this study as well. Positive pacing strategies with a quicker pace at the beginning segments of the race can often be explained as a result of "pre-race enthusiasm" or current absence of fatigue [4]. As a result, fatigue is more likely to occur later in race, thus reducing the running speed [8].
In all age groups, marathon runners (both men and women) showed significantly greater PaceR than half-marathoners. More even pacing in the half-marathon than the marathon indicates that aging plays a more important role as the race distance increases [19]. This phenomenon could exist due to the occurrence of severe fatigue in the marathon race after the 35th kilometer, which is often caused by glycogen depletion [11] rather than some psychological factors (e.g., fast race start due to competitiveness) [6]. However, the psychological characteristics of younger male runners (particularly younger than 30) could result in more variability in pacing. This variability appears primarily as a result of a fast start induced by their higher levels of self-esteem, which results in overestimation of their performance capabilities [26]. Moreover, younger, less experienced long distance runners could face problems with the control mechanism responsible for pacing, thus changing pace more often, which can result in rapid increase of fatigue. The important factors in setting an overall pacing strategy for a bout of exercise include knowledge of the endpoint and the associated duration of the event, an internal clock using scalar timing, and the memory of pacing strategy from prior events [3]. Younger runners with insufficient experience, could lack this "pacing template" in the brain, hence they change pace more often. On the other hand, elderly men spend more time running (i.e., run slowly). As a consequence, fatigue as well as increased pacing variability is more likely to occur.
Pacing in the women's marathon and half-marathon does not seem to be influenced by age, which can be attributed to a better fatigue tolerance, regardless of age. Women generally have more fatigue-resistant type I muscle fibers, and are able to better utilize fat when participating in long distance running [14,35]. In addition, women are less likely to start a race at high-speed. The rationale for this can be found in women's motivation for running long distance races. They are less competitive than men, and their motivation is to socialize, enjoy the race, and eventually, cross the finish line [33,36]. Nevertheless, the trend observed in Figure 2, showed that young and elderly women do have greater PaceR, particularly in marathons (however it was not statistically significant). The aforementioned findings could help sports and medicine practitioners to create separate age specific training plans and pacing strategies for men and women. This approach could help long distance runners to improve their physical fitness. Consequently, they could achieve better race times, reduce the potential risk of musculoskeletal injuries and increase the overall pleasure of long distance running.
Contrary to our previous study involving the Ljubljana half-marathon [19], an end spurt was observed in the Vienna marathon and half-marathon races. Race configuration, altitude, weather conditions, as well as the number of participants are possible explanations for these diametrically opposite outcomes. As a result, further studies of end spurts in marathons, and particularly half-marathons are required. In essence, the occurrence of a final spurt might be due to psychological factors. We have stated earlier, that pacing is a combination of anticipation, knowledge of the end-point, prior experience and an internal clock using scalar timing [3,37]. The knowledge of the near finish might motivate the runners to mobilize the last reserves. In particular, younger half-marathon men and women runners demonstrated the faster end spurts compared to the older age groups and marathon runners. These results further support our observation that younger long distance runners produce more variability in pacing. As a result, the end spurt was as much as 3.83% faster than Section 4 (in half-marathon women). Sports and medicine practitioners could use these findings to further familiarize young long distance runners with the most optimal pacing strategies, with the aim of enhancing performance, as well as reducing the risk of cardio-vascular emergencies, which are often induced by a fast end-spurt or high pace variability. Finally, race organizers and official race medical services could also benefit from these findings. Namely, they can spread awareness of the negative consequences of fast end spurts and high pace variability to race participants, thus preventing possible medical emergencies.
A limited number of participants in age groups older than 60 years prevented inclusion of older age groups and therefore limits our knowledge about pacing strategies in older runners. Furthermore, the absence of additional factors that could influence pacing, such as running experience, previous training routines, runners' personal characteristics, running in large groups for social interaction, or limited weather information, can be considered as study limitations. Finally, this study has included only one event held in one year (i.e., the 2017 Vienna City Marathon), thus limiting the potential generalization of the findings. Conversely, the greatest strength of this study was its novelty, since it adds original information to the existing literature regarding one of the most popular race distances.

Conclusions
To conclude, in both men and women regardless of age group, marathon runners showed greater variability in pacing than half-marathon runners. In addition, women showed no differences in pace variability with regard to age group whereas the youngest and oldest men showed the greatest variability in pace. Finally, younger half-marathon men and women runners showed the fastest end spurts compared to older age groups and marathon runners. The presented findings could help sports and medicine practitioners to create separate age specific training plans and pacing strategies for men and women. This approach could help long distance runners to adopt an even or negative pacing profile with a low to moderate end spurt. This pacing approach could improve physical fitness, help runners to achieve better race times, reduce the potential risk of musculoskeletal injuries, and increase the overall pleasure in long distance running.