Tower Running—Participation, Performance Trends, and Sex Difference

hough there are exhaustive data about participation, performance trends, and sex di erences in performance in di erent running disciplines and races, no study has analyzed these trends in stair climbing and tower running. The aim of the present study was therefore to investigate these trends in tower running. The data, consisting of 28,203 observations from 24,007 climbers between 2014 and 2019, were analyzed. The e ects of sex and age, together with the tower characteristics (i.e., stairs and oors), were examined through a multivariable statistical model with random e ects on intercept, at climber's level, accounting for repeated measurements. Men were faster than women in each age group (p<0.001 for ages 69 years,p=0.003 for ages>69 years), and the di erence in performance stayed around 0.20 km/h, with a minimum of 0.17 at the oldest age. However, women were able to outperform men in speci c situations: (i) in smaller buildings (<600 stairs), for ages between 30 and 59 years and>69 years; (ii) in higher buildings (>2200 stairs), for age groups<20 years and 60–69 years; and (iii) in buildings with 1600–2200 stairs, for ages>69 years. In summary, men were faster than women in this speci c running discipline; however, women were able to outperform men in very speci c situations (i.e., speci c age groups and speci c numbers of stairs). Keywords:tower running; sex di erences; age; running speed; vertical run 1. Introduction Distance running is

(iii) in buildings with 1600–2200 stairs, for ages>69 years. In summary, men were faster than women in this speci c running discipline; however, women were able to outperform men in very speci c situations (i.e., speci c age groups and speci c numbers of stairs). Keywords:tower running; sex di erences; age; running speed; vertical run 1. Introduction Distance running is of high popularity and includes di erent distances, from 5 to 10 km [1], half-marathon [2,3], marathon [2,4], and up to ultra-marathon of di erent distances [5,6]. It is well-known that men are faster than women from 5 km to marathon [7], and in ultra-marathon running [8]. However, women were able to reduce the gap with men in ultra-marathon running, with increasing age and at longer race distances [9]. Stair climbing or tower running is a very speci c running discipline, in which stair climbing has developed into the organized sport of tower running. Nowadays, tower running is a sport discipline that involves running up tall buildings, such as internal staircases of skyscrapers. However, tower running can cover any running race that involves a course that ascends a building. To date, we have knowledge about the health bene ts of stair climbing [10–13]. However, no data exist about participation and performance trends in tower running, and especially about the sex di erence in this speci c running discipline. Such information is valuable for athletes and coaches, Int. J. Environ. Res. Public Health2020,17, 1902; doi:10.3390 /ijerph17061902 /journal/ijerph

Int. J. Environ. Res. Public Health2020,17, 1902 2 of 9 to better understand and plan a race strategy, and also for race organizers, for insights regarding future events. Therefore, the aim of the present study was to investigate participation and performance trends in tower running, with the hypothesis that men would also be faster than women in this discipline. Regarding age groups, we expected that women might close the performance gap in the older groups as already shown in long distance races [9]. 2. Materials and Methods 2.1. Ethics 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. 2.2. Methodology There exists a tower running world association that presents all the results of the known races around the world on their homepage (www.towerrunning.com). In an older version of this homepage, there were only the results of the current year, and sometimes, of the preceding year. We contacted the person in charge at the association to nd out whether he could provide us with older data as well. For some races, however, it was not possible to nd the older results. For example, for the race at the Willis tower in Chicago, the results before 2018 were not available. For other races, such as the hustle up race in Chicago, the direct link did not work, but the results could be found by searching for the link to the race, which is also provided on the homepage of the tower running world association. Table summarizes all considered events listed by the number of steps of the buildings. Table 1.Data included in the present study. Building City Steps Data Available (Years) Included (Years) Millennium Tower Wien 2529 2014–2016 2014–2016 Willis Tower (Sears Tower until 2009) Chicago 2109 2014–2019 2018 Taipei 101 Taipeh 2046 2014–2019 2017–2018 CN Tower Toronto 1776 2014–2019 2017–2018 Reunion Tower Dallas 1674 2018–2019 2018 Ei elturm Paris 1665 2015–2020 2015–2018 AON Center Chicago 1643 none on towerrunning.com 2018 John Hancock Center

study. Building City Steps Data Available (Years) Included (Years) Millennium Tower Wien 2529 2014–2016 2014–2016 Willis Tower (Sears Tower until 2009) Chicago 2109 2014–2019 2018 Taipei 101 Taipeh 2046 2014–2019 2017–2018 CN Tower Toronto 1776 2014–2019 2017–2018 Reunion Tower Dallas 1674 2018–2019 2018 Ei elturm Paris 1665 2015–2020 2015–2018 AON Center Chicago 1643 none on towerrunning.com 2018 John Hancock Center (875 North Michigan Avenue) Chicago 1632 2014–2019 2017–2018 Empire State Building New York 1576 2014–2019 2017–2014 Bank of America Plaza Dallas 1540 none on towerrunning.com 2018 US Bank Tower Los Angeles 1500 2014–2019 2018 thyssenkrupp Testturm Rottweil 1390 2018–2019 2018 Swissætel The Stamford Singapur 1336 2014–2018 2017 Rockefeller Center New York City 1214 2014–2016, 2018, 2019 2019 MesseTurm Frankfurt am Main 1202 2014–2019 2014–2017 Three Logan Square Philadelphia 1088 2014–2019 2014, 2018, 2019 Valliance Bank Oklahoma City 837 2014–2019 2019 Holmenkollbakken Oslo 800 2015–2018 2015–2017 Run Up Berlin (Park Inn Hotel) Berlin 770 2015–2019 2015–2018 KölnTurm Köln 714 none on towerrunning.com 2016–2019 Oakbrook Terrace Tower Oakbrook 680 2014–2020 2019 Münsterturm Ulm 560 none on towerrunning.com 2014–2018 Towerrun Berlin 465 2014–2020 2018 St.George's Tower Leicester 351 none on towerrunning.com 2018 Matzleinsdorfer Hochhaus Wien 342 2017 2017 Windradlauf Lichtenegg 300 2014 2014 Haus des Meeres Wien 271 2015–2019 2016–2018 Oluempia Hotel Tallinn N /A 2015–2019 2017

Int. J. Environ. Res. Public Health2020,17, 1902 3 of 9 From the race results, the year of the event, the completed time, the sex, and the name of both the athletes and the building were available. We further looked for the height of the building and the number of stairs and oors. Race time in m:sec was converted to running speed in km/h, using the height of the building. We removed observations from unknown climbers (where the name of the climber was not reported or not known) in order to correctly account for repeated measurements. We also considered multi-climbing. 2.3. Statistical Analysis The outcome was the tower climbing speed (km/h). Descriptive statistics are presented as means (SD=standard deviations) by sex and age groups. T-tests were performed to assess the outcome di erence between sex, overall and for each age groups. Two-way ANOVA tests were also performed to evaluate the multivariable e ect of sex and age on the outcome. Then, to control also for repeated measurements and the other covariates, the e ects of sex and age, together with the tower characteristics (i.e., stairs and oors) were examined more rigorously through a multivariable mixed e ects model, with random e ects (intercept) for climbers. The model was speci ed as follows: Tower climbing speed (Y) ~ [Fixed e ects (X)=Sex*Age*BS (Stairs, df=5) +BS (Floors, df=5)+[random e ects of intercept=runners] where BS (Stairs, df=5) and BS (Floors, df=5) are 5 degrees of freedom (df) basis splines changing with the number of stairs and oors, respectively; Sex*Age*BS (Stairs, df=5) denoted the three-way interaction term Sex–Age–number of stairs. Calendar year was not considered in the above model because it was not signi cant. Results of the regression model are presented as estimates and standard errors. Statistical significance was defined asp<0.05. All statistical analyses were carried out with R, R Core Team (2016). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria ( /foundation/). The R packages ggplot2, lme4, and lmerTest were used, respectively, for data visualization and for the mixed model. The R code to reproduce the

standard errors. Statistical significance was defined asp<0.05. All statistical analyses were carried out with R, R Core Team (2016). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria ( /foundation/). The R packages ggplot2, lme4, and lmerTest were used, respectively, for data visualization and for the mixed model. The R code to reproduce the analysis is provided as supplementary information (Supplement 1 R-code). 3. Results Between 2014 and 2019, the total number of observations was 28,203 (24,007 climbers). However, the total number of observations, with non-missing sex, was 28,156 (23,960 climbers). The participation and men-to-women ratio is shown in Figure. We observed that we had a low number of participants and a high men-to-women ratio before 2017 (i.e., the number of men was three times the number of women in 2015). The highest number of participants was recorded in 2018. In fact, the number of women in 2018 was eight times the number of women in 2014, and the number of men in 2018 was four times the number of men in 2014. In 2019, the number of available observations decreased again. The men-to-women ratio reached a minimum in 2019 with 0.89, meaning that the number of women was higher than the number of men. In Table, the mean performance by sex and age group is reported. Men were faster than women in each age group (p<0.001 for all ages until 69 years,p=0.003 for ages>69 years), and the di erence in performance stayed around 0.20 km/h, with a minimum of 0.17 at the oldest age. In Table, summary statistics of performance, together with tower characteristics: height, number of oors and stairs are reported by sex. Overall, the sex di erence in performance was signi cant (p<0.001); sex di erences were also signi cant (p<0.001) in average oors and stairs climbed. The results of the multivariable statistical analysis are displayed in Figure, to allow an easier interpretation and understanding. Moreover, we had no signi cant di erence between men and women alone, but in the interaction with age groups and stairs climbed (Supplemental 2 Table).

cant (p<0.001); sex di erences were also signi cant (p<0.001) in average oors and stairs climbed. The results of the multivariable statistical analysis are displayed in Figure, to allow an easier interpretation and understanding. Moreover, we had no signi cant di erence between men and women alone, but in the interaction with age groups and stairs climbed (Supplemental 2 Table). The variability, in terms of performance, was greater in very young and very old age groups (<20 years, 60–69 years, and>69 years). This also

Int. J. Environ. Res. Public Health2020,17, 1902 4 of 9 had an e ect on sex di erences. Women performed better than men in the following situations: (i) smaller buildings (<600 stairs), for ages between 30 and 59 years and>69 years; (ii) higher buildings (>2200 stairs), for age<20 years and ages between 60 and 69 years; and (iii) buildings with 1600 to 2200 stairs, for age>69 years. In all other cases, men performed better than women, with the sex di erence reducing when the number of stairs increased. In Figure, the e ect of the number of oors on performance, by sex, is shown. When the number of oors increased, the average speed of tower climbing decreased, but then increased around 90 oors, and decreased again in climbing the highest buildings.Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 4 of 9 the number of women in 2015). The highest number of participants was recorded in 2018. In fact, the number of women in 2018 was eight times the number of women in 2014, and the number of men in 2018 was four times the number of men in 2014. In 2019, the number of available observations decreased again. The men-to-women ratio reached a minimum in 2019 with 0.89, meaning that the number of women was higher than the number of men. Figure 1. Participation and men-to-women ratio. In Table 2, the mean performance by sex and age group is reported. Men were faster than women in each age group (p<0.001 for all ages until 69 years, p=0.003 for ages >69 years), and the difference in performance stayed around 0.20 km/h, with a minimum of 0.17 at the oldest age. In Table 3, summary statistics of performance, together with tower characteristics: height, number of floors and stairs are reported by sex. Overall, the sex difference in performance was significant (p<0.001); sex differences were also significant (p<0.001) in average floors and stairs climbed. The results of the multivariable statistical analysis are displayed in Figure 2, to allow an easier interpretation and understanding. Moreover, we had no significant difference

together with tower characteristics: height, number of floors and stairs are reported by sex. Overall, the sex difference in performance was significant (p<0.001); sex differences were also significant (p<0.001) in average floors and stairs climbed. The results of the multivariable statistical analysis are displayed in Figure 2, to allow an easier interpretation and understanding. Moreover, we had no significant difference between men and women alone, but in the interaction with age groups and stairs climbed (Supplemental 2 Table). The variability, in terms of performance, was greater in very young and very old age groups (<20 years, 60–69 years, and >69 years). This also had an effect on sex differences. Women performed better than men in the following situations: (i) smaller buildings (<600 stairs), for ages between 30 and 59 years and >69 years; (ii) higher buildings (>2200 stairs), for age <20 years and ages between 60 and 69 years; and (iii) buildings with 1600 to 2200 stairs, for age >69 years. In all other cases, men performed better than women, with the sex difference reducing when the number of stairs increased. In Figure 3, the effect of the number of floors on performance, by sex, is shown. When the number of floors increased, the average speed of tower climbing decreased, but then increased around 90 floors, and decreased again in climbing the highest buildings. Figure 1.Participation and men-to-women ratio. Table 2. Summary statistics of tower climbing performance, running speed (km/h), by sex and age groups.p-values from t-tests for each subgroup are reported.p-values from ANOVA were bothp<0.001 for sex and age. Men-to-women ratio, computed with the number of participants, is reported. Age Group Sex N Mean (SD) p Men-to-Women Ratio <20 F 501 0.73 (0.27) <0.001 1.30 M 652 0.91 (0.38) 20–29 F 1887 0.81 (0.24) <0.001 1.39 M 2615 0.99 (0.35) 30–39 F 2552 0.80 (0.30) <0.001 1.34 M 3415 1.03 (0.39) 40–49 F 1941 0.78 (0.32) <0.001 1.33 M 2583 1.00 (0.39) 50–59 F 1220 0.76 (0.30) <0.001 1.60 M 1951 0.97 (0.38) 60–69 F 239 0.72 (0.25) <0.001 2.62 M 626 0.90 (0.27) >69 F 44 0.66 (0.33)

20–29 F 1887 0.81 (0.24) <0.001 1.39 M 2615 0.99 (0.35) 30–39 F 2552 0.80 (0.30) <0.001 1.34 M 3415 1.03 (0.39) 40–49 F 1941 0.78 (0.32) <0.001 1.33 M 2583 1.00 (0.39) 50–59 F 1220 0.76 (0.30) <0.001 1.60 M 1951 0.97 (0.38) 60–69 F 239 0.72 (0.25) <0.001 2.62 M 626 0.90 (0.27) >69 F 44 0.66 (0.33) 0.003 4.57 M 201 0.83 (0.33)

Int. J. Environ. Res. Public Health2020,17, 1902 5 of 9Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 6 of 9 Figure 2. Speed (km/h) by stairs, age, and sex. Lines represent the predicted values from the mixed model and points represent the average of the observed values. Figure 3. Speed (km/h) by floors and sex. Lines represent the predicted values from the mixed model and points represent the average of the observed values. Figure 2. Speed (km/h) by stairs, age, and sex. Lines represent the predicted values from the mixed model and points represent the average of the observed values.Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 6 of 9 Figure 2. Speed (km/h) by stairs, age, and sex. Lines represent the predicted values from the mixed model and points represent the average of the observed values. Figure 3. Speed (km/h) by floors and sex. Lines represent the predicted values from the mixed model and points represent the average of the observed values. Figure 3. Speed (km/h) by oors and sex. Lines represent the predicted values from the mixed model and points represent the average of the observed values.

Int. J. Environ. Res. Public Health2020,17, 1902 6 of 9 Table 3. Summary statistics of running speed (km/h) and race time (min), tower height (m), oors, and stairs by sex. Data expressed as mean ( SD). Females (n=11,886) Males (n=16,270)p-Value Speed km/h 0.85 (0.37) 1.06 (0.46) <0.001 Time (min) 24.26 (14.16) 18.43 (11.69) <0.001 Tower height (m) 296.25 (111.37) 276.27 (108.72) <0.001 Floors 85.44 (36.37) 76.00 (35.97) <0.001 Stairs 1466.43 (420.36) 1401.18 (429.59) <0.001 4. Discussion The aim of the present study was to investigate participation trends, performance trends, and trends in sex di erence in tower running, with the hypothesis that men would be faster than women in this discipline. The main ndings were: (1) more men than women competed before 2017, (2) men were faster than women in each age group and the di erence in performance stayed around 0.20 km/h, with a minimum of 0.17 km/h at the oldest age, and (3) women aged between 30 and 59 years and>69 years performed better than men in smaller buildings (<600 stairs). 4.1. Change in the Men-to-Women Ratio Across Years Before 2017, we observed a low number of participants and a high men-to-women ratio. The highest number of participants was recorded in 2018. In 2019, the number of participants decreased again and the men-to-women ratio reached the minimum of 0.89, which means that the number of women was higher than the number of men. This could also be due to a selection bias. At the time of the data collection (2017–2019), there were more results available from the earlier races and since the aim of the selection was to represent the sport and include the most important races all over the world, we did not pay attention to compare for every year the exact same number of races. This fact should encourage race directors to join the `Towerrunning World Association' (www.towerrunning.com), in order to build up a rm data base for future analyses. Generally, in races of long traditions, the men-to-women ratio is>1.0, indicating that more men than women competed [14], but the men-to-women ratio can decrease over the

The study analyzes trends in tower running performance and participation, focusing on sex differences.