The prediction of running performances is very important for athletes and trainers (1–3). Predictions might be used to detect talents, select a speed for specific pace training sessions, and even evaluate the effect of a training program (1–4). Sports researchers have therefore long sought ways to accurately predict running performances. Several authors have developed certain methods and tools (e.g., power law, critical speed or nomograms) (5–8). These different tools are used to predict middle- and long-distance track-running performances, but only in non-obstacle races. Consequently, the studies on predicting the track-running performances of the steeplechase remain very limited.

The 3,000-m steeplechase is a technical middle-distance running event with the athletes crossing 28 barriers and 7 barriers followed by a water pit (water-jumps). The main technical/regulatory difference between the sexes is the height of the barriers (i.e., 0.914 m and 0.762 m for men and women, respectively).

Some predictive variables of a 3,000-m steeplechase might be the same variables as for flat middle- and long-distance running. Maximal oxygen uptake (V˙O_2max), endurance capacity or running economy are key physiological factors of running performances (9). These factors known to be dependent on age, sex, height, body mass and/or body mass index (BMI) (10, 11). For age, few studies have investigated it relationship with performance among steeplechase runners. One study has nevertheless shown that the age of elite athletes at peak performance on the 3,000-m steeplechase is around 25–27 years old (12). Sex also appears to be a variable to consider in predicting performance in the 3,000-m steeplechase. Differences in the technique used to cross the steeplechase barriers have been identified between the sexes (13), probably due to the lower barrier height for women. Therefore, the same equation cannot be used for men and women. These differences in hurdling technique may also be due to the approach velocity and body height (14). Sánchez Muñoz et al. (15) showed that runners of long-distance events (including the 3,000-m steeplechase) are taller than runners of shorter distances (800-m and 1,500-m). Being taller is often associated with longer legs, which would facilitate barrier crossing. Moreover, it is well established that a lower BMI positively impacts running performances (16). Consequently, due to the technical aspect and the energy cost of crossing barriers (17), height, body mass and BMI appear to be other predictive variables of steeplechase performance.

The purpose of the current study was to develop and test the validity and accuracy of a simple equation in predicting 3,000-m steeplechase performance (Perf_Steeple) from individual variables (e.g., age, height, body mass and BMI) in men.

The characteristics of the athletes in both groups are presented in Table 1.

Significant bivariate correlations were found between Perf₃₀₀₀ (p < 0.001, r = 0.896), age (p < 0.001, r = 0.474), body mass (p = 0.049, r = 0.154), BMI (p = 0.001, r = 0.279) and Perf_Steeple.

Stepwise multiple linear regression analysis using Perf₃₀₀₀ and age as potential independent variables yielded the simple prediction equation.

This equation was as follows:PerfSteeple=−57,165+1,147×Perf3000+0,955×agePerf_Steeple= −57,165 + 1,147 ×

Perf_Steeple= −57,165 + 1,147 ×  Perf₃₀₀₀ + 0,955 × age

with Perf_Steeple and Perf₃₀₀₀ in seconds, and age in years.

VIF was 1.206 for both Perf₃₀₀₀ and age.

The increases in r² by adding a second predictor (i.e., age) into the prediction equation were significant with F(1,114) = 8.083 (p = 0.005). Moreover, Fisher's test revealed a p < 0.05.

The performance estimated by the simple prediction equation (including the 2 independent variables: Perf₃₀₀₀ and age) was significantly correlated with the actual Perf_Steeple (r = 0.903 and r² = 0.8162).

The standardized β coefficients and p-values on the Student's t-test were 0.844 (p < 0.001), and 0.126 (p = 0.005) for Perf₃₀₀₀ and age, respectively.

The Durbin-Watson test indicated no autocorrelation in the residuals, with a value of 1.725. The standardized residuals followed a straight line, showing that no residual was too high, so the prediction was valid for all performances.

No significant difference was noted between the actual and predicted Perf_Steeple (p = 0.263), and the magnitude of the difference was considered trivial (ES = –.08) (22).

Predicted Perf_Steeple was significantly correlated with actual Perf_Steeple (p < 0.001). A very high correlation coefficient was found according to Munro's scale (23) (r = 0.929, Figure 1).

The bias and 95% LoA were −5 ± 24 s, i.e., −0.8 ± 7.6% of mean Perf_Steeple (Figure 1). In 95 of 100 new predictions, the difference between actual and predicted performance would be less or equal to −5 ± 24 s.

The purpose of this study was to develop and test the validity and accuracy of a simple equation to predict Perf_Steeple. The main result suggests that a simple equation (i.e., Perf_Steeple = −57.165 + 1.147 × Perf₃₀₀₀ + 0.955 × age) is a valid and accurate method for predicting Perf_Steeple. Although no study has focused on predicting Perf_Steeple from a simple equation, a few studies have validated predictive methods for the flat 3,000-m with results comparable (i.e., from low bias and 95% LoA) to our own (3, 27). This confirms the validity of the present simple equation for the 3,000-m steeplechase. This simple equation can complement other methods/tools. A long-distance runner (who wanted to predict Perf_Steeple without reference in Perf₃₀₀₀) could use the methods and tools available in the literature to firstly estimate Perf₃₀₀₀ (5–8). Next, use this estimated Perf₃₀₀₀ to predict his Perf_Steeple with the simple equation.

The present results suggest that the main predictor of Perf_Steeple is Perf₃₀₀₀. The duration in these two events is relatively close (i.e., 8%–10% prolonged effort in the 3,000-m steeplechase, having little impact on the importance of the V˙O_2max) (17, 28, 29). The 3,000-m steeplechase generally produces a deceleration during the takeoff step for both hurdling and water-jumps (30). In addition to requiring a high level of V˙O_2max (29), Perf_Steeple requires a high technical ability to cross the barriers. This high technical ability (e.g., more efficiency over barriers) might be developed by plyometric training (17). The inclusion of plyometric training could support runners to be more efficient over barriers (17). These types of exercises increased muscular fitness which can be useful to decelerating, landing and exiting steeplechase barriers (31).

Based on the statistical method used (20, 24, 26), this study shows that age contributes to performance prediction in the Perf_Steeple. Its inclusion provides only a small gain in accuracy. It can be explained by the observation that most steeplechasers are between 20 and 35 years old (Table 1). This is not surprising, given that a peak age of around 25 years old has been noted for men (12). However, the inclusion of age could be explained by the age-related physiological modifications with adult aging (32). A progressive reduction in V˙O_2max appears to be the primary mechanism associated with performance declines (33).

The current study suggests that height, body mass and BMI are not predictive variables of Perf_Steeple. This result might be explained by the inclusion of Perf₃₀₀₀ in the equation. Although height is slightly associated with speed in running events, body mass and BMI seems to be a better indicator of Perf₃₀₀₀ (16). The impact of height, body mass and BMI on Perf₃₀₀₀ would explain why these variables were not included in the equation (i.e., they are already included in Perf₃₀₀₀ and would therefore be redundant).

The main limitation of this study is that the running performances were not executed in similar conditions (i.e., environmental, physical and psychological). Another limitation was the self-declaration of body height and mass and thus the calculation of the BMI. However, runners are known to self-report accurately for this type of data (34). Finally, this equation is a valid and accurate method only in male athletes. The height of the barriers with a 0.152-m difference between sexes, can lead to differences in the running-hurdling (13). The lower barrier height relative to body height for women might help them to maintain race pace when crossing the barriers (14). However, they are more affected by the water-jump (e.g., lower approach velocity, shorter jump) than those of the men (13). This is probably because they need a longer stride to cross the water due to their generally shorter leg length. Thus, a future perspective might be to develop a specific prediction equation to predict Perf_Steeple for women.

The results of the present study indicate the validity and accuracy of a simple equation to predict Perf_Steeple. This tool could be useful for athletes and coaches to predict a 3,000-m steeplechase. It can enable them to adapt their training and to select the optimum speed for the athlete's performance. It can be use for national directors in charge of performance in athletics federations to detect talent in the 3,000-m steeplechase and thus orient young athletes more easily.