Effects of 12-week cadence retraining on impact peak, load rates and lower extremity biomechanics in running

Participants

Sample size estimation indicated that a minimum sample size of 26 participants was required to achieve a minimum effect size of 0.6. Considering a drop-out rate of 15–20%, 30 male recreational runners were recruited through online social media, running clubs and flyers. Participants were randomly assigned to either a cadence retraining group (CAD) or a control group (CON) on the basis of the lottery method of sampling, and 15 participants were included in each group (Table 1). When the participants first reported to the laboratory, they were required to run on a treadmill for 2 min. A high-speed camera placed next to the treadmill recorded their foot strike patterns. The participants were determined to be rearfoot strikers by checking the foot angle (i.e., the angle between the foot and ground at initial contact) of the dominant leg (the preferred kicking leg) (Fu et al., 2017) by reviewing the obtained videos frame by frame (Heiderscheit et al., 2011). Consequently, all runners were rearfoot strikers. They ran a minimum of 15 km/week for at least 3 months prior to the study. Participants were excluded if they had any lower limb musculoskeletal injuries in the previous 6 months. No significant differences in age, height, weight and weekly mileage were observed between the two groups. This study was approved by the Institutional Review Board of the Shanghai University of Sports (No. 2017007). Informed written consent was obtained from each participant prior to their participation in this study.

Instrumentation

A 12-camera motion capture system (100 Hz, T40; Vicon Motion Inc., Oxford, UK) was used to collect kinematic data. Ground reaction force data were captured by using two 90 cm × 60 cm × 10 cm force platforms (1,000 Hz, 9287B; Kistler Instruments AG Corp., Winterthur, Switzerland). The kinematics and ground reaction force data were simultaneously collected using the Vicon system. A Photogate system (Witty-Wireless Training Timer; Microgate Corp., Bolzano, Italy) was used to monitor over-ground running speed. Conventional running shoes (Nike Air Zoom Pegasus 34) were used by the participants during the experiments (Fig. 1A).

Experimental protocol

The participants visited the laboratory twice, at baseline and at the end of the gait retraining program. Prior to data collection, the participants were required to wear uniform clothing, including a vest, socks and shoes, and walk for 2 min and run at 3.33 m/s on a treadmill for 5 min as a warm-up. Thereafter, a total of 40 markers were placed on the participants, and static calibration was performed. The anatomical locations of the markers were the right/left ilium crest tubercle, right/left posterior superior iliac spine, right/left femur greater trochanter, right/left anterior superior iliac spine, right/left femur lateral epicondyle, right/left femur medial epicondyle, right/left fibula apex of the lateral malleolus, right/left tibia apex of the medial malleolus, right/left head of the fifth metatarsals, right/left head of the first metatarsus and right/left posterior surface of the calcaneus (Fig. 1B). In addition, three tracking markers were placed on the thigh and shank. The participants were instructed to run over the ground across a 10 m runway (Fig. 1C) at 3.33 m/s during which kinematic and ground reaction force data were captured. The running speed was considered acceptable if the deviation was within 5%. Three successful running trials were collected for each participant.

Retraining protocol

All of the participants were required to run at their preferred speeds during the cadence retraining protocol (Hafer et al., 2015). Running speed and cadence during training were monitored using the commercial running application CODOON^© (Chengdu Ledong Information Technology Co., Ltd., Chengdu, China). Each participant received a sport belt bag in which to place their mobile phones during running, and they were instructed to place the bag above their sacrum. The participants were asked to run outdoors three times (30 min/run) at a comfortable speed to determine their preferred speed and cadence. The preferred speed and preferred cadence were the average values obtained from three outdoor trials. Participants in the CAD group were instructed to run with a 7.5% increase in cadence, whereas those in the CON group ran without any change in cadence (Willy et al., 2016b). Participants in the retraining group were informed about and given access to a mobile-based metronome application with tempos set to a 7.5% increase in cadence. Figure 2 shows the cadence retraining protocol, which lasted for 12 weeks with three sessions a week and 5–48 min each session (Neal et al., 2018). Participants used their preferred running mode, namely, treadmill or over ground, to complete their retraining. The retraining protocol constituted part of the participants’ running volume so that their total weekly running volume remained unchanged. After each retraining session, participants could check their average cadence, speed and running volume on the CODOON^© running application. They were also required to submit data recorded by the application to the researchers. Participants were excluded if their training protocols were interrupted more than three times or if their cadence did not achieve the targeted cadence for 3 weeks since the beginning of training. Weekly group trainings were provided three times a week in the CAD group to ensure compliance, and participants chose one of weekly group training sessions in which to participate on the basis of their schedule. During group training, the participants performed an 8 min warm-up, such as dynamic stretching, under the guidance of the researchers. Then, the participants began to run according to their retraining schedule. The participants did not receive guidance on running techniques because the weekly group training aimed to ensure compliance and the quality of retraining.

Data processing

The Visual 3D software (v5; C-Motion, Inc., Germantown, MD, USA) was used to compute the 3D kinematic and kinetic variables of the lower extremity during running. Marker trajectories were filtered with a cut-off frequency of 7 Hz via a fourth-order Butterworth low-pass filter (Yang et al., 2020). A seven-segment lower extremity model was built via the Visual 3D, and CoG was estimated from this model. Impact force variables included impact peak, vertical instantaneous load rates (VILR) and vertical average load rates (VALR). In rearfoot strike runners, impact peak was defined the first peak in the ground reaction force curve (Fig. 3B). Load rates was calculated on the basis of the method described by Futrell et al. (2018). In brief, a point of interest (POI) was defined as the first point above 75% of a participant’s body weight with an instantaneous load rate of less than 15 body weight/s. VALR (the average slope) and VILR (i.e., the maximum instantaneous slope) were then calculated from 20% to 80% and from 20% to 100% of the force at POI, respectively (Fig. 3B). We also calculated lower extremity stiffness (Liu et al., 2006), k_leg, as shown in Eq. (1). Kinematic variables of the hip, knee and ankle joints included foot angle (the angle between the foot and ground) at initial contact (Fig. 3A) and peak joint extension and peak joint flexion angles during the stance. The times from initial contact to impact peak (t_ip), vertical velocities of the CoG at initial contact and vertical excursion of the CoG during the stance phase were also evaluated.

(1) $$k_{\mathrm{l}\mathrm{e}\mathrm{g}}={\displaystyle \frac{\mathrm{G}\mathrm{R}{\mathrm{F}}_{\mathrm{i}}}{\mathrm{\Delta }y}}$$where GRF_i is the vertical ground reaction force at the lowest position of the CoG and Δy is the maximum vertical displacement of the CoG.

Statistics

The mean and standard deviation for each variable was calculated. Two-way repeated measure ANOVA was used to characterise the effects of training (pre- and post-training) and group (CAD and CON) on each variable. Independent sample and paired t-tests were used as post-hoc tests when a significant interaction was detected to assess potential group effects between CAD and CON and retraining effects pre- and post-training, respectively. The observed effect size (η²) was considered in the ANOVA results, and effect size (Cohen’s d) was considered in the paired and independent sample t-tests results. The 95% confidence interval (CI) of the differences in group effects was reported. The criterion α level was set to 0.05. All statistical procedures were conducted using SPSS software (Version 20; SPSS, Inc., Chicago, IL, USA).

Dropout rate

Thirty participants (15 in the CAD group and 15 in the CON group) completed the pre-training tests on their first visit to the laboratory (Table 1). However, in the CAD group, one participant was excluded because of insufficient training volume, and two participants withdrew for personal reasons or because their results showed more than three interruptions. In the CON group, two participants were lost to contact, and one participant withdrew for personal reasons. Overall, 24 participants, 12 in the CAD group and 12 in the CON group, completed the 12-week cadence retraining protocol and reported to the laboratory for post-training tests (Table 1). No significant difference in average running volumes was observed between the CAD and CON groups (CAD: 23.3 ± 3.3 km/week, CON: 22.9 ± 4.3 km/week).

Cadence and step length

Figure 4A shows a significant training × group interaction effect for cadence (P < 0.001, η² = 0.867). Specifically, cadence significantly increased by 5.7% (161.3 ± 9.5 step/min vs. 170.5 ± 9.2 step/min) in the CAD group (P < 0.001, Cohen’s d = 3.87) but not in the CON group (164.3 ± 7.7 step/min vs. 165.5 ± 6.8 step/min, P > 0.05) after training. A significant main effect of training was observed for step length, which decreased by 4.1% (2.49 ± 0.16 m vs. 2.39 ± 0.14 m) in the CAD group after training (P = 0.011, η² = 0.259) (Fig. 4B). No significant difference for step length was observed in the CON group (2.54 ± 0.16 m vs. 2.51 ± 0.14 m). Step length in the CAD group was 4.9% lower (2.39 ± 0.14 m vs. 2.51 ± 0.14 m) than that in the CON group after training (P = 0.04, 95% CI [−0.245 to −0.006], Cohen’s d = 0.94).

Impact force variables

Figure 5A shows significant training × group interaction effects for impact peak (P = 0.022, η² = 0.217). Impact peak significantly decreased in the CAD group (P = 0.003, Cohen’s d = 1.10) but not in the CON group (P > 0.05). Meanwhile, impact peak in the CAD group was significantly lower than that in the CON group after training (P = 0.038, 95% CI [−0.443 to −0.013], Cohen’s d = 0.95). Significant main effects of training were observed for VALR and VILR. Specifically, VALR (P = 0.029, η² = 0.198) and VILR (P = 0.025, η² = 0.209) decreased in the CAD group after training (Table 2).

Kinematics and joint mechanics

Significant training × group interaction effects were observed for foot angle (P = 0.04, η² = 0.178), vertical velocity of the CoG at initial contact (P = 0.035, η² = 0.186) and vertical excursion of the CoG (P = 0.001, η² = 0.409). Foot angle (P = 0.003, Cohen’s d = 1.09), vertical velocity of the CoG at initial contact (P = 0.002, Cohen’s d = 1.16) and vertical excursion of the CoG (P < 0.001, Cohen’s d = 1.83) decreased in the CAD group after training (Table 2; Fig. 5). Moreover, vertical excursion of the CoG in the CAD group was significantly lower than that in the CON group after training (P = 0.025, 95% CI [−0.015 to −0.001], Cohen’s d = 1.03). Significant main effects of training were observed for peak knee flexion angle. Specifically, peak knee flexion angle (P = 0.048, η² = 0.166) was decreased in the CAD group after training (Table 2). A significant main effect of training was observed for lower extremity stiffness, which increased in the CAD group after training (P = 0.048, η² = 0.166) (Table 2).