Effects of 16 weeks of plyometric training on knee biomechanics during the landing phase in athletes

Abstract

This study investigated the effects of plyometric training on lower‐limb muscle strength and knee biomechanical characteristics during the landing phase. Twenty‐four male subjects were recruited for this study with a randomised controlled design. They were randomly divided into a plyometric training group and a traditional training group and underwent training for 16 weeks. Each subject was evaluated every 8 weeks for knee and hip isokinetic muscle strength as well as knee kinematics and kinetics during landing. The results indicated significant group and time interaction effects for knee extension strength (F = 74.942 and p = 0.001), hip extension strength (F = 99.763 and p = 0.000) and hip flexion strength (F = 182.922 and p = 0.000). For landing kinematics, there were significant group main effects for knee flexion angle range (F = 4.429 and p = 0.047), significant time main effects for valgus angle (F = 6.502 and p = 0.011) and significant group and time interaction effects for internal rotation angle range (F = 5.475 and p = 0.008). The group main effect for maximum knee flexion angle was significant (F = 7.534 and p = 0.012), and the group and time interaction effect for maximum internal rotation angle was significant (F = 15.737 and p = 0.001). For landing kinetics, the group main effect of the loading rate was significant (F = 4.576 and p = 0.044). Significant group and time interaction effects were observed for knee extension moment at the moment of maximum vertical ground reaction force (F = 5.095 and p = 0.010) and for abduction moment (F = 8.250 and p = 0.001). These findings suggest that plyometric training leads to greater improvements in hip and knee muscle strength and beneficial changes in knee biomechanics during landing compared to traditional training.

Highlights

• Plyometric training changes landing biomechanics.

• Changes help to reduce the risk of ACL injury.

• Changes from long‐term training are more noticeable.

1. INTRODUCTION

Incident anterior cruciate ligament (ACL) injuries that occur in athletes absent of direct contact with the external environment are often referred to as non‐contact ACL injuries (Owoeye et al., 2020). Notably, landing from a height has been considered one of the high‐risk manoeuvres resulting in these injuries. Deficits in lower‐limb strength during landing are an important risk factor for non‐contact ACL injuries. Such deficits may lead to decreased dynamic knee stability during landing, failure to effectively inhibit dynamic knee valgus (Sasaki et al., 2019) and rotation (Shultz et al., 2015) and an increased risk of non‐contact ACL injury (Yu et al., 2007). Meanwhile, resistance training is a traditional and effective means of increasing lower extremity muscle strength. Enhanced lower extremity strength has the potential to modify biomechanical risk factors associated with ACL injury during landing. McCurdy et al. (2012) performed short‐term lower‐limb resistance training on 13 adult females and found that resistance training was able to increase the angle of knee flexion and decrease joint stiffness during landing in the subjects compared with the control group. However, a significant amount of research suggests that besides traditional resistance training, training in neuromuscular control appears to be more effective in reducing the risk of ACL injury (Ahmadabadi et al., 2023; Emery et al., 2022; Zebis et al., 2016). Plyometric training is a type of training that emphasises stretched contraction cycles, which are generated by eccentric contractions followed by rapid concentric contractions, to increase neuromuscular recruitment and increase maximal output force (Grgic et al., 2021).

Plyometric training is a typical form of neuromuscular training which can modulate a variety of receptors, such as the muscle spindle and Golgi tendon organs, and change the pattern of neuromuscular activity during exercise (Hewett et al., 2005). Several studies have applied this method to the prevention of lower limb sports injuries with positive results. Weltin et al. (2017) found that 4 weeks of plyometric training can improve trunk control and reduce knee valgus angle and abduction moments during the execution of lateral cuts compared with a control group, thus reducing the risk of knee injury. Similarly, Brown et al. (Tsang et al., 2011) indicated that 6 weeks of plyometric training changed the biomechanical characteristics of the knee joint during athletes' two‐legged landings, which reduced the risk of knee injuries.

By contrast, several studies have suggested that plyometric training has little effect on the prevention of knee injuries during landing. As Dos' Santos et al. (Dos’Santos et al., 2019) stated in their review that plyometric training did not change the knee kinetics in the change of direction manoeuvres. Ruan et al. (2022) also found that athletes receiving plyometric training interventions did not show significant changes in biomechanical characteristics during landing. Moreover, studies that have reported significant changes in landing biomechanics following plyometric training tended to be less robust in evidence. Most of these studies primarily documented changes in a single plane of motion such as the coronal or sagittal planes. Consequently, supporting evidence is needed to demonstrate whether this type of training is effective in reducing the risk of ACL injury during landing considering that the mechanics of this injury in the knee are potentially multi‐planar with a combined loading pattern (Berns et al., 1992; Markolf et al., 1995). Meanwhile, the majority of available studies have focussed on the effects of short‐term (4–8 weeks) plyometric training (Ahmadabadi et al., 2023; Brown et al., 2014; Cherni et al., 2019; Ghanati et al., 2022; Weltin et al., 2017; Zebis et al., 2016). By contrast, few studies have examined the long‐term effects of such training on the biomechanical characteristics of the knee during landing (>12 weeks) (Wang et al., 2023). Thus, the long‐term effects of plyometric training on the biomechanical characteristics of landing require further investigation.

The present study performed 16 weeks of plyometric training in male college basketball players and evaluated the effects of plyometric training on lower‐limb muscle strength and three‐dimensional kinematic and kinetic characteristics of the knee during landing with an isometric plyometric testing system as well as a three‐dimensional infrared motion capture system. This research aimed to investigate whether plyometric training could change the biomechanical dangerous characteristics associated with non‐contact ACL injuries during the landing of college basketball players compared with traditional training. This study was expected to provide additional insights into the theoretical and practical references for the prevention of ACL injuries in the knee joint.

2. METHODS

2.1. Participants

The G*Power 3.0.10 programme was used to calculate the minimum sample size needed for our study. We based the calculation on an expected effect size (f) of 0.25, which is considered a medium effect size and a desired power (1–β err prob) of 0.7 to detect significant differences with a confidence level of 95% (α = 0.05). We set the number of groups to 2 and the number of measurements to 3. As a result, the analysis determined that a minimum of 12 participants were required for both the experimental and control groups to achieve sufficient statistical power. Twenty‐four male basketball players were recruited as experimental subjects in this study. All participants had at least 2 years of basketball training experience, competed in the China University Basketball League (CUBA) and were ranked in the top 48 teams in the league. They engaged in regular basketball training sessions averaging 10 h per week. Random number table method was used to divide the subjects randomly into two groups: the traditional training group (n = 12) and the plyometric training group (n = 12). The traditional strength training used muscular resistance exercises on the lower limbs without plyometric component, and plyometric training consisted mainly of various forms of jumping training.

The inclusion criteria for this study were (i) male, member of the university basketball team; (ii) dominant side is right side; (iii) no history of nerve, muscle or bone injury of the lower limbs in the last 1 year; (iv) stable mental state, no psychiatric or psychological illness and (v) voluntary participation in this study and commitment to participate in the training and assessment on time.

Drop‐out criteria for this study were (i) subjects were unwilling to participate in this study again and (ii) subjects had injuries during the training cycle, which interfered with training and assessment (e.g., ankle fracture, anterior cruciate ligament tear and meniscus tear).

The physical characteristics of the subjects are shown in Table 1.

TABLE 1.

Physical characteristics of subjects.

	Traditional training	Plyometric training
Age (years)	21.71 ± 3.21	21.52 ± 2.89
Height (m)	1.94 ± 6.93	1.92 ± 8.24
Weight (kg)	87.46 ± 5.39	88.74 ± 6.38
BMI (kg/m2)	23.24 ± 3.22	24.07 ± 4.13
Training experience (mouth)	25.50 ± 2.24	25.75 ± 1.98
Weekly training (h)	10	10

2.2. Experimental procedure

In the present study, resistance training and plyometric training were conducted separately in two groups for 16 weeks. The plyometric training was divided into three phases: the first phase was an adaptation phase with a moderate intensity for 4 weeks; the second phase was an enhancement phase with a medium‐high intensity for 4 weeks and the third phase was a consolidation phase with a high intensity for 8 weeks. The training cycle for traditional training was congruent with plyometric training, and the variations in training intensity were consistent with plyometric training. The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Ethical Review Board: SUDA20211227H03. All participants signed a written informed consent form.

Baseline tests of physical characteristics, strength and landing biomechanics were performed on all subjects before the start of the experiment. Isometric muscle strength tests and landing manoeuvre biomechanics tests were performed on subjects at weeks 8 and 16. The baseline tests included height, weight, leg length, knee width and ankle width. Strength tests included 1RM of barbell squat, hexagonal bar deadlift, Bulgarian split squat, lunge squat and weighted calf raise and isokinetic plyometrics with 120°/s angular velocity. The landing manoeuvre biomechanics were tested as the subject dropped from a 50 cm jump using a two‐legged landing with kinetic features collected using a three‐dimensional force table and kinematic features collected using a high‐speed infrared camera. Only isometric plyometric tests and landing manoeuvre biomechanics tests were repeated at the weeks 8 and 16 of training.

2.2.1. Isokinetic strength testing

Isometric muscle strength of the knee and hip joints at 120°/s was tested using the CON‐TREX isometric strength testing and training system (Physiomed, Germany). This system provides high precision with a measurement accuracy of ±0.5% of the full scale and an output resolution of 0.1 N.

After the warm‐up of a 10‐min self‐selected intensity bike ride, isometric muscle strength testing of the knee was performed. Subjects were seated on the isokinetic strength testing system with the center of the knee joint coinciding with the fixed end of the machine's rotating arm. The range of motion in the test was the angle at which each subject felt that they were able to generate the most muscle force by their own subjective perception. The procedure for isometric muscle strength testing of the hip was the same as for the knee.

Three tests were administered to each subject, with a three‐minute break between each test. The maximum peak torque produced was taken as the final evaluation. Verbal encouragement was given during the test to stimulate the subject to produce maximum torque. Limbs were weighed prior to each test according to the instructions in the dynamometer manual to exclude the effects of limb gravity. Isokinetic peak torques (N·m·kg⁻¹) normalised by body weight were used for statistical analysis.

Peak torque was the indicator.

2.2.2. Landing biomechanical tests

Biomechanical characteristics in the sagittal, coronal and horizontal planes of the subjects during landing were captured using a Vicon three‐dimensional motion capture system. This system consists of 8 high‐resolution infrared cameras (model: MX13, Vicon Motion Analysis, UK), each with a resolution of 13 megapixels. The camera sampling frequency was set to 100 Hz, ensuring sub‐millimetre accuracy in capturing motion data.

Ground reaction forces were measured using two Kistler three‐dimensional force plates (model: 9281, Kistler Instrumente AG, Switzerland) placed side by side. Each force plate measures 60 × 40 × 10 cm and has a sampling frequency of 1000 Hz. The precision of the force plates is 0.1 N, ensuring highly accurate measurement of ground reaction forces during landing.

Subjects wore tight‐fitting clothing and were introduced to the relevant test apparatus as well as the test procedure to familiarise them with the test environment. After the standard procedure of warming up, the staff affixed marked dots to the subject's lower limbs for modeling the lower limbs. To minimise experimental errors, all pasting was done by one person only.

Then, the Vicon workstation was activated to record the biomechanical data. The subject stood on the force table remaining stationary with their arms spread out to capture a static model. After static acquisition, the biomechanical characteristics of the lower limbs for landing manoeuvres were started. The landing manoeuvre test protocol featured the subject standing with both legs on a step, which was followed by the dominant side lower limb extending out of the step. After the staff signaled the start of the test, the subject left the steps and landed both feet in the center of each of the two force tables. During the test, the subjects maintained their hands at waist level with no arm swing. Each subject performed three landing tests, and the results of the second test were selected as the final assessment. Should the results of the second assessment be unavailable, the results of the third or first test were selected for analysis. The indexes included: knee X‐axis, Y‐axis and Z‐axis range of motion (°); X‐axis, Y‐axis and Z‐axis maximum angle (°); load rate (N/s) and X‐axis, Y‐axis and Z‐axis maximum moments (N·m/kg).

2.2.3. Training protocols

The traditional training group and the plyometric training group each trained concurrently for 16 weeks, four times per week, for a total of approximately 64 training sessions per group each. Each session lasted approximately 90 min, and all training was guided and supervised by a trained physical trainer. A standard warm‐up procedure was completed by both groups before the start of each training session, which consisted of 5 minutes of self‐dynamic stretching and 10 min of jogging. Post‐training recovery measures consisted of jogging and self‐active stretching. This consisted of 20 min of jogging and 30 s of active stretching of each of the major muscles of the lower limbs.

The traditional training group performed with barbell squats, hexagonal bar deadlift, Bulgarian split squat, forward lunge, reverse lunge and weighted calf raise. Moderate intensity was set for training in weeks 1–4 with moderate intensity defined as 60%–70% 1RM. Moderate–high intensity was set for training in weeks 5–8 with moderate–high intensity defined as 70%–80% 1RM. High intensity was set for training in weeks 9–16 with high intensity defined as 70%–90% 1RM.

The plyometric training group trained using various forms of jumping such as long jumps, vertical jumps and deep jumps. Similarly, moderate intensity training was employed during weeks 1–4, which consisted of vertical jumps and box jumps, with subjects performing approximately 100 touches per training session. Moderate–high intensity training was employed during weeks 5–8, which consisted of consecutive jumps and box jumps, with subjects performing approximately 120 touches per training session. High‐intensity training was employed during weeks 9–16, which consisted of box jumps, with subjects performing approximately 100 touches per training session. The specific protocols are shown in Table 2.

TABLE 2.

Training protocols.

Type	Phases	Exercise	Intensity	Repetitions and sets
Plyometric training	1–4 weeks	Tuck jump	Medium	10 × 2
		Long jump	Low	10 × 2
		1/4 squat jump	Low	10 × 2
		Box jump (40 cm)	Low	10 × 2
		Box jump down (40 cm)	Medium	10 × 2
	5–8 weeks	CM hurdle jump	Medium	6 × 4
		Continuous long jump	Medium	6 × 4
		Box jump without arm swing (40 cm)	Medium	6 × 4
		Lateral box jump (40 cm)	Medium	6 × 4
		Drop jump (40 cm)	High	6 × 4
	9–16 weeks	Lateral hurdle jump	Medium	10 × 2
		Depth to box jump (40 cm)	High	5 × 4
		Depth to broad jump (40 cm)	High	5 × 4
		Lateral box jump (40 cm)	High	5 × 4
		Rotational vertical hop	Medium	10 × 2
Traditional training	1–4 weeks	Barbell squat	70% 1RM	6 × 3
		Hexagonal bar deadlift	70% 1RM	6 × 3
		Bulgarian split squat	60% 1RM	10 × 3
		Forward lunge	60% 1RM	10 × 3
		Reverse lunge	60% 1RM	10 × 3
		Weighted calf raise	60% 1RM	10 × 3
	5–8 weeks	Barbell squat	80% 1RM	6 × 3
		Hexagonal bar deadlift	80% 1RM	6 × 3
		Bulgarian split squat	70% 1RM	10 × 3
		Forward lunge	70% 1RM	10 × 3
		Reverse lunge	70% 1RM	10 × 3
		Weighted calf raise	70% 1RM	10 × 3
	8–16 weeks	Barbell squat	90% 1RM	6 × 3
		Hexagonal bar deadlift	90% 1RM	6 × 3
		Bulgarian split squat	70% 1RM	10 × 3
		Forward lunge	70% 1RM	10 × 3
		Reverse lunge	70% 1RM	10 × 3
		Weighted calf raise	70% 1RM	10 × 3

Both training protocols were based on the NSCA‐CSCS Essentials of Strength Training and Conditioning (4th Edition). To reduce potential biases due to different training modalities, we matched the overall training volume and intensity between groups. Both groups trained four times per week for 16 weeks under the supervision of qualified coaches. Training sessions were adjusted to equate for perceived exertion, ensuring comparable training stimuli despite different exercise types.

2.3. Data statistics

The kinematic trajectories of the reflective labeled balls captured by Vicon were named and censored using Nexus software (version 2.7.1). Then, the kinematic and kinetic data were imported into the Visual3D (C‐Motion, version 5.02.3) software. A static skeletal model with nine links was created from the static file. Low‐pass filtering was performed in Visual3D with a cutoff frequency distribution of 6 and 20 Hz (second‐order Butterworth). Three‐dimensional angle changes of the knee joint were obtained by transforming with Cardan angle, and three‐dimensional moments of the knee joint were calculated with inverse dynamics algorithm. The X‐axis direction was defined as knee extension (−)/flexion (+), the Y‐axis direction was the knee adduction (−)/abduction (+) and the Z‐axis direction was the knee external rotation (−)/internal rotation (+). Since the joint moments in this study were calculated from the ground reaction forces by an inverse kinetics algorithm, all the joint moments in this study are external moments.

The moment that the subject touched the ground was defined as the first time when the value of the vertical ground reaction force (vGRF) was not zero. The moment of maximum knee flexion angle was defined as the time when the knee angle of movement on the X‐axis was maximum. The period of landing motion was defined from the moment the subject touched the ground to the moment of maximum knee flexion angle.

The knee X, Y and Z‐axis moments for each subject were normalised using body weight. The calculation of loading rate (LR) is shown in Equation (1). vGRF means the peak vGRF, and t1–t0 means the time from the touchdown instant to the peak vGRF during the landing of subjects. Peak isometric muscle torques were normalised by body weight as well (Pamukoff et al., 2016).

$$\text{LR}=\frac{\text{vGRF}}{\begin{array}{c}t1-t0\end{array}},$$ (1)

$$\text{Landing}\text{Rate}$$

Statistical analyses were performed using SPSS 26.0, and the results of the experiments were expressed as M ± SD. The Shapiro–Wilk test was used to test whether the experimental data obeyed normal distribution and Levene's test for data variance chi‐square. Data that conformed to a normal distribution and passed the variance chi‐square test were examined using repeated measures ANOVA. Simple effects analyses were conducted if the interaction effect between time and group was significant. Main effects analyses were conducted if the interaction effect was not significant. Generalised estimating equations were used to test the biased data. The level of significance was α = 0.05.

Effect sizes were calculated to quantify the magnitude of differences observed. For ANOVA, partial eta squared (ηp²) was reported as the effect size measure. ηp² value of 0.01 indicates a small effect, 0.06 indicates a medium effect and 0.14 indicates a large effect.

3. RESULTS

3.1. Isokinetic muscle strength test results

The data obeyed a normal distribution by the Shapiro–Wilk test.

3.1.1. Knee extensor torque

The knee extensor torque data matched the test of sphericity (Mauchly's W = 0.918 and p = 0.290).

Group main effect: There was a significant main effect for the group, indicating that overall knee extensor torque was significantly different between plyometric and traditional training groups.

Time main effect: There was a significant main effect for time, indicating that knee extensor torque changed significantly over the training period.

Group × time interaction effect: The ANOVA results indicated a significant group × time interaction effect for knee extensor torque. Simple effects analysis showed no significant difference in knee extension torques between the two groups before training (p = 0.901). However, knee extension torques were significantly higher in the plyometric training group compared to the traditional training group at week 8 (p = 0.001, 95% CI [0.291 and 0.397]) and week 16 (p = 0.000, 95% CI [0.300 and 0.410]).

3.1.2. Knee flexion torque

The knee flexion torque data did not conform to the test of sphericity (Mauchly's W = 0.628 and p = 0.001) and used the Greenhouse–Geisser results.

Group main effect: The main effect for the group was not significant, indicating no overall difference in knee flexion torque between the plyometric and traditional training groups.

Time main effect: There was a significant main effect for time, indicating that knee flexion torque changed significantly over the training period.

Group × time interaction effect: The interaction effect between time and group was not significant.

The results are shown in Table 3 and Figure 1.

TABLE 3.

Isokinetic strength of knee.

	Group	Baseline (N·m/kg)	8 weeks (N·m/kg)	16 weeks (N·m/kg)	Repeated measures ANOVA
					F	p	ηp2
Extension	Traditional training	1.72 ± 0.16	1.81 ± 0.17	1.89 ± 0.17
	Plyometric training	1.72 ± 0.17	2.16 ± 0.21	2.24 ± 0.19
	Group main effect				195.772	0.000	0.871
	Time main effect				237.944	0.000	0.893
	Interactive effect				74.942	0.001	0.710
Flexion	Traditional training	1.52 ± 0.32	1.62 ± 0.25	1.68 ± 0.19
	Plyometric training	1.51 ± 0.36	1.63 ± 0.31	1.70 ± 0.23
	Group main effect				0.008	0.930	0.000
	Time main effect				113.779	0.000	0.791
	Interactive effect				1.151	0.311	0.037

FIGURE 1.

Results of isokinetic muscle strength.

3.1.3. Hip extensor torque

The hip extensor torque data met the test of sphericity (Mauchly's W = 0.927 and p = 0.332).

Group main effect: There was a significant main effect for the group, indicating that overall hip extensor torque was significantly different between the plyometric and traditional training groups.

Time main effect: There was a significant main effect for time, indicating that hip extensor torque changed significantly over the training period.

Group × time interaction effect: The ANOVA results indicated a significant group × time interaction effect for hip extensor torque. Simple effects analysis showed no significant difference in hip extension torques between the two groups before training (p = 0.580). However, hip extension torques were significantly higher in the plyometric training group compared to the traditional training group at week 8 (p = 0.001, 95% CI [0.052 and 0.183]) and week 16 (p = 0.000, 95% CI [0.266 and 0.401]).

3.1.4. Hip flexor torque

The hip flexor torque data met the test of sphericity (Mauchly's W = 0.934 and p = 0.372).

Group main effect: There was a significant main effect for the group, indicating that overall hip flexor torque was significantly different between the plyometric and traditional training groups.

Time main effect: There was a significant main effect for time, indicating that hip flexor torque changed significantly over the training period.

Group × time interaction effect: The ANOVA results indicated a significant group × time interaction effect for hip flexor torque. Simple effects analysis showed no significant difference in hip flexion torques between the two groups before training (p = 0.581). However, hip flexion torques were significantly higher in the plyometric training group compared to the traditional training group at week 8 (p = 0.000, 95% CI [0.051 and 0.154]) and week 16 (p = 0.000, 95% CI [0.297 and 0.423]). Specific results are shown in Table 4 and Figure 1.

TABLE 4.

Isokinetic strength of hip.

	Group	Baseline (N·m/kg)	8 weeks (N·m/kg)	16 weeks (N·m/kg)	Repeated measures ANOVA
					F	p	ηp2
Extension	Traditional training	2.17 ± 0.08	2.21 ± 0.09	2.36 ± 0.11
	Plyometric training	2.15 ± 0.09	2.32 ± 0.09	2.70 ± 0.10
	Group main effect				25.397	0.000	0.458
	Time main effect				472.289	0.000	0.940
	Interactive effect				99.763	0.000	0.769
Flexion	Traditional training	1.91 ± 0.17	1.96 ± 0.08	2.09 ± 0.08
	Plyometric training	1.90 ± 0.04	2.06 ± 0.06	2.45 ± 0.10
	Group main effect				40.446	0.000	0.574
	Time main effect				619.382	0.000	0.950
	Interactive effect				182.922	0.000	0.859

3.2. Kinematic characteristics of knee

3.2.1. Range of knee angle changes

The data obeyed a normal distribution by the Shapiro‐Wilk test.

Knee flexion angle

The knee flexion angle data met the test of sphericity (Mauchly's W = 0.808 and p = 0.107).

Group main effect: There was a significant main effect for the group, indicating that the knee flexion angle differed significantly between the groups overall.

Time main effect: There was a significant main effect for time, indicating significant changes in knee flexion angle over time.

Group × time interaction effect: The group × time interaction effect was not significant.

Group main effects analysis revealed no significant difference between the two groups before training (p = 0.599, 95% CI [−2.776 and 4.669]) and at week 8 (p = 0.325, 95% CI [−11.293 and 3.916]). However, the plyometric training group presented a significantly higher knee flexion angle than the traditional training group at week 16 (p = 0.011, 95% CI [−14.915 and −2.199]).

Knee valgus angle

The knee valgus angle data met the test of sphericity (Mauchly's W = 0.872 and p = 0.239).

Group main effect: The main effect for the group was not significant, indicating no overall difference in knee valgus angle between the groups.

Time main effect: The time main effect was significant, indicating significant changes in knee valgus angle over time.

Group × time interaction effect: The group × time interaction effect was not significant.

Knee internal rotation angle

The knee internal rotation angle data met the test of sphericity (Mauchly's W = 0.927 and p = 0.332).

Group main effect: There was a significant main effect for the group, indicating that the knee internal rotation angle differed significantly between the groups overall.

Time main effect: The time main effect was not significant.

Group × time interaction effect: The ANOVA results indicated a significant group × time interaction effect for knee internal rotation angle.

The results of the simple effects analysis indicated no significant difference in the knee rotation angle between the traditional and plyometric training groups before training (p = 0.826) and at week 8 of training (p = 0.239). However, the knee rotation angle in the plyometric training group was significantly lower at week 16 compared to the traditional training group (p = 0.000, 95% CI [−15.306 and −7.174]).

The specific results are shown in Table 5.

TABLE 5.

Range of knee angles during landing.

	Group	Baseline (°)	8 weeks (°)	16 weeks (°)	Repeated measures ANOVA
					F	p	ηp2
Flexion angle	Traditional training	64.56 ± 5.30	65.24 ± 8.28	66.25 ± 7.33
	Plyometric training	63.60 ± 3.29	68.93 ± 9.63	74.80 ± 7.68
	Group main effect				4.429	0.047	0.168
	Time main effect				5.044	0.011	0.187
	Interactive effect				2.745	0.075	0.111
Valgus angle	Traditional training	14.20 ± 4.05	8.28 ± 4.06	16.31 ± 7.15
	Plyometric training	15.36 ± 8.56	11.25 ± 6.75	15.18 ± 5.68
	Group main effect				0.436	0.516	0.019
	Time main effect				6.502	0.003	0.228
	Interactive effect				0.664	0.520	0.029
Internal rotation angle	Traditional training	24.59 ± 6.93	25.38 ± 7.63	26.45 ± 5.32
	Plyometric training	24.11 ± 2.65	21.87 ± 6.53	15.21 ± 4.22
	Group main effect				13.831	0.001	0.386
	Time main effect				2.457	0.097	0.100
	Interactive effect				5.475	0.008	0.199

3.2.2. Maximum knee angle the data obeyed a normal distribution by the Shapiro–Wilk test

Maximum knee flexion angle

The maximum knee flexion angle data met the test of sphericity (Mauchly's W = 0.928 and p = 0.456).

Group main effect: There was a significant main effect for the group, indicating that the maximum knee flexion angle differed significantly between the groups overall.

Time main effect: There was a significant main effect for time, indicating significant changes in maximum knee flexion angle over time.

Group × Time Interaction Effect: The group × time interaction effect was not significant.

Group main effects analysis indicated no significant difference in the maximal flexion angle between the two groups before training (p = 0.712). The maximal flexion angle was significantly larger in the plyometric training group than in the conventional training group at week 8 (p = 0.003, 95% CI [2.028 and 8.336]) and week 16 (p = 0.000, 95% CI [1.650 and 4.698]) of training.

Maximum knee valgus angle

The maximum knee valgus angle data met the test of sphericity (Mauchly's W = 0.823 and p = 0.130).

Group main effect: The main effect for the group was not significant, indicating no overall difference in the maximum knee valgus angle between the groups.

Time Main Effect: The time main effect was not significant.

Group × Time Interaction Effect: The group × time interaction effect was not significant.

Maximum knee internal rotation angle

The maximum knee internal rotation angle data met the test of sphericity (Mauchly's W = 0.900 and p = 0.331).

Group main effect: There was a significant main effect for the group, indicating that the maximum knee internal rotation angle differed significantly between the groups overall.

Time main effect: There was a significant main effect for time, indicating significant changes in the maximum knee internal rotation angle over time.

Group × time interaction effect: The group × time interaction effect was significant.

The results of the simple effects analysis indicated no significant difference in the maximum internal rotation angle between the two groups before training (p = 0.985). The maximum internal rotation angle was significantly greater in the plyometric training group than in the conventional training group at week 8 (p = 0.012, 95% CI [0.540 and 3.943]) and week 16 (p = 0.000, 95% CI [6.250 and 9.009]) of training.

Specific results are shown in Table 6 and Figure 2.

TABLE 6.

Maximum knee angle during landing.

	Group	Baseline (°)	8 weeks (°)	16 weeks (°)	Repeated measures ANOVA
					F	p	ηp2
Flexion angle	Traditional training	−73.12 ± 6.86	78.83 ± 3.34	84.89 ± 1.97
	Plyometric training	−73.99 ± 4.32	84.01 ± 4.08	88.06 ± 1.62
	Group main effect				7.534	0.012	0.255
	Time main effect				74.427	0.000	0.772
	Interactive effect				2.040	0.142	0.085
Valgus angle	Traditional training	25.42 ± 2.94	25.03 ± 7.64	24.63 ± 5.06
	Plyometric training	25.00 ± 5.08	22.64 ± 7.31	24.30 ± 3.41
	Group main effect				0.461	0.504	0.021
	Time main effect				0.466	0.631	0.021
	Interactive effect				0.331	0.720	0.015
Internal rotation angle	Traditional training	25.96 ± 2.43	24.33 ± 1.94	26.98 ± 1.65
	Plyometric training	25.83 ± 3.14	22.09 ± 2.08	19.35 ± 1.61
	Group main effect				67.627	0.000	0.755
	Time main effect				10.446	0.000	0.322
	Interactive effect				15.737	0.001	0.417

FIGURE 2.

Maximum knee angle during landing.

3.3. Kinetic characteristics of knee

3.3.1. Loading rate

The data obeyed a normal distribution by the Shapiro–Wilk test and matched a spherical test (Mauchly's W = 0.756 and p = 0.053).

Group main effect: There was a significant main effect for the group, indicating that the loading rate differed significantly between the groups overall.

Time main effect: There was a significant main effect for time, indicating significant changes in the loading rate over time.

Group × Time Interaction Effect: The group × time interaction effect was not significant.

Group main effects analysis indicated no significant difference in the loading rates between the two groups before training (p = 0.412, 95% CI [−3.153 and 7.417]). The loading rates in the plyometric training group were significantly lower than in the traditional training group at week 8 (p = 0.007, 95% CI [1.396 and 7.698]) and week 16 (p = 0.025, 95% CI [0.549 and 7.394]).

The specific results are shown in Table 7.

TABLE 7.

Results of loading rate in landing.

Group	Baseline (N/s)	8 weeks (N/s)	16 weeks (N/s)	Repeated measures ANOVA
				F	p	ηp2
Traditional training	39.73 ± 7.97	24.86 ± 3.96	21.38 ± 4.45
Plyometric training	37.60 ± 4.80	21.14 ± 4.01	18.41 ± 4.77
Group main effect				4.576	0.044	0.172
Time main effect				115.246	0.000	0.840
Interactive effect				0.178	0.677	0.008

3.3.2. Maximum moments of knee

The data obeyed a normal distribution by the Shapiro‐Wilk test.

Knee extension moment

The extension moment data did not meet the test of sphericity (Mauchly's W = 0.707 and p = 0.026), thus Greenhouse–Geisser corrections were applied.

Group main effect: There was a significant main effect for the group, indicating differences in knee extension moments between the groups overall.

Time main effect: The time main effect was significant, indicating significant changes in knee extension moments over time.

Group × Time Interaction Effect: The group × time interaction effect was significant.

The results of the simple effects analysis indicated no significant difference existed in extension moments between the two groups before training (p = 0.943). However, knee extension moments were significantly lower in the plyometric training group than in the traditional training group at week 8 (p = 0.005, 95% CI [0.077 and 0.374]) and week 16 (p = 0.001, 95% CI [0.171 and 0.536]) of training.

Knee abduction moment

The abduction moment data did not meet the test for sphericity (Mauchly's W = 0.751 and p = 0.049), thus Greenhouse–Geisser corrections were applied.

Group main effect: The main effect for the group was not significant, indicating no overall difference in abduction moments between the groups.

Time main effect: The time main effect was significant, indicating significant changes in abduction moments over time.

Group × Time Interaction Effect: The group × time interaction effect was not significant.

Knee internal rotation moment

The internal rotation moment data did not meet the test of sphericity (Mauchly's W = 0.510 and p = 0.001), thus Greenhouse–Geisser corrections were applied.

Group main effect: There was a significant main effect for the group, indicating differences in internal rotation moments between the groups overall.

Time main effect: The time main effect was significant, indicating significant changes in internal rotation moments over time.

Group × Time Interaction Effect: The group × time interaction effect was significant.

The results of the simple effects analysis indicated no significant difference existed in rotational moments between the plyometric training group and the traditional training group before training (p = 0.558) and at week 8 of training (p = 0.084). However, rotational moments were significantly lower in the plyometric training group than in the traditional training group at week 16 (p = 0.000, 95% CI [−0.216 and −0.079]).

Specific results are shown in Table 8 and Figure 3.

TABLE 8.

Maximum knee moment in landing.

	Group	Baseline (N·m/kg)	8 weeks (N·m/kg)	16 weeks (N·m/kg)	Repeated measures ANOVA
					F	p	ηp2
Extension moment	Traditional training	−1.68 ± 0.31	−1.42 ± 0.19	−1.17 ± 0.12
	Plyometric training	−1.69 ± 0.30	−1.20 ± 0.15	−0.82 ± 0.28
	Group main effect				11.136	0.003	0.336
	Time main effect				52.612	0.000	0.705
	Interactive effect				3.695	0.046	0.144
Abduction moments	Traditional training	0.52 ± 0.12	0.41 ± 0.12	0.34 ± 0.10
	Plyometric training	0.60 ± 0.18	0.36 ± 0.08	0.27 ± 0.10
	Group main effect				0.266	0.611	0.012
	Time main effect				43.871	0.000	0.666
	Interactive effect				3.105	0.068	0.124
Internal rotation moment	Traditional training	0.76 ± 0.11	0.69 ± 0.12	0.59 ± 0.09
	Plyometric training	0.74 ± 0.08	0.60 ± 0.14	0.44 ± 0.07
	Group main effect				7.053	0.014	0.243
	Time main effect				52.251	0.000	0.704
	Interactive effect				3.852	0.048	0.149

FIGURE 3.

Maximum knee moment in landing.

4. DISCUSSION

This study performed 16 weeks of plyometric training and traditional strength training on two groups of college basketball players to compare the differences in strength of the lower limbs and biomechanical characteristics of the knee joint during landing between these two types of training. Distinct from previous studies, the present study extended the training period to 16 weeks to assess the long‐term effects of plyometric training.

We found that traditional strength training and plyometric training improved lower‐limb muscle strength and changed knee kinematics and kinetics characteristics during landing. However, compared with conventional training, plyometric training was more effective in increasing hip extension, hip flexion and knee extension muscle torque for subjects. Additionally, in terms of knee biomechanics during landing, plyometric training was able to increase the knee flexion angle, decrease the rotation angle and decrease the knee loading rate and moment more significantly.

Results of isometric muscle strength indicated that traditional training and plyometric training improved hip and knee muscle torque at 120°/s. The plyometric training group also showed more distinct increases in the isometric torque of the hip extensors, hip flexors and knee extensors than the traditional strength training group. By contrast, the change in isometric torque of the knee flexors was not significantly different between the two groups. Considerable research has suggested that plyometric training can improve the strength of the knee extensor muscles. Lephart et al. (2005) found that plyometric training was effective in increasing knee extension strength with no significant difference in flexion strength compared with the control group. The finding is consistent with the present study and is confirmed by a number of studies (Cherni et al., 2019; Krishna et al., 2019; Myer, Ford, Brent, & Hewett, 2006). The technical movements of plyometric training are based on jumping, which results in less intense stimulation of the flexor muscles and less pronounced strength gains in the flexor muscles than in the extensor muscles. Ghanati et al. (2022) observed the effect of neuromuscular training on hip strength in athletes and found that plyometric training was able to increase hip abduction isometric torque in athletes. Hip muscle strength affects knee biomechanical characteristics during landing to some extent as well, which makes hip muscle strength worthy of attention.

Both trainings in this study changed the landing biomechanical characteristics in the sagittal plane of the knee. The maximum knee flexion angle increased during landing in both groups of athletes, while the maximum extension moment tended to decrease. In between‐group comparisons, we found that joint angles and joint moments in the sagittal plane during landing were significantly different. Specifically, the plyometric training group had a larger knee flexion angle and a lower extension moment. Previous studies have demonstrated that these changes can reduce the risk of non‐contact ACL injuries (Gonzalez‐Jurado et al., 2016). For example, it has been reported that increasing the knee flexion angle can reduce anterior displacement of the tibia when a tensile force is applied, thereby lowering the risk of ACL injury (Engebretsen et al., 2012). Furthermore, research indicates that the quadriceps force arm on the anterior tibial pull is greater at lower knee flexion angles, resulting in increased anterior shear forces at the knee joint (DeMorat et al., 2004). Plyometric training has been associated with reduced extension moments in athletes, suggesting lower sagittal plane stresses on the knee, which in turn diminishes the risk of ACL injury (Sell et al., 2007; Zebis et al., 2009). Additionally, plyometric training has been shown to enhance muscle strength and neuromuscular control of the knee extension and hip extension muscles (Hewett et al., 2005). This increase in muscle strength prolongs the landing cushioning time, leading to greater knee flexion angles and reduced extension moments (King et al., 2021).

Several changes occurred in the biomechanical characteristics of the coronal plane after training as well. Although, neither traditional nor plyometric training changed the kinematic characteristics of the coronal plane, they reduced the knee abduction moment. Furthermore, athletes with plyometric training had significantly lower knee abduction moments in landing than the conventionally trained group, which might be related to the increased cushioning time caused by the increased knee flexion angle. With increased cushioning time, the ground reaction force was generated in landing decreases, thus resulting in less stress on the knee joint (Yu et al., 2006) and a reduction in the abduction moment. Video studies have noted that most ACL injuries are accompanied by a large knee valgus angle, thus suggesting a link between a large valgus angle and a high risk of ACL injury during landing (Koga et al., 2010). We noted that athletes with plyometric training in the study by Weltin et al. (2017) effectively reduced dynamic knee valgus in landing. Differences in this outcome might be attributed to differences in training protocols. In Weltin et al.’s study, the training protocol used plyometric training for core strength, whereas the present study focussed on lower‐limb strength improvement. This difference might indirectly suggest that core strength and core control could be more helpful in inhibiting a dynamic knee valgus during landing. Besides kinematic characteristics, changes in kinetics are one of factors impacting the risk of ACL injury. Miyasaka et al. (2002) found that an increase in knee abduction moment could increase ACL pressure within the 0°–90° flexion angle. Fleming et al. (2001) obtained similar results and found that the knee abduction moment in the weight‐bearing condition would increase the pressure on the ACL.

The knee internal rotation angle and internal rotation moment in the horizontal plane is one of the main risk factors for ACL injury as well (Oh et al., 2012). Fleming et al. (2001) indicated that knee internal rotation moments in a load‐bearing condition significantly increased the pressure on the ACL. The plyometric training was effective in reducing the knee internal rotation angle and internal rotation moment compared with conventional training in the present study. A decrease in the knee internal rotation angle and the internal rotation moment appears to be associated with the improvement of balance in the studies available. Cochrane et al. (2010) indicated that subjects with balance training had significantly lower internal rotation moments during cutting manoeuvres. Li et al. (2018) also indicated that patients with chronic ankle instability presented high knee internal rotation moments in landing. Moreover, Giesche et al. (2021) conducted a meta‐analysis and found a significant increase in knee internal rotation moments when subjects performed unplanned movement tasks. These results might indicate that the balance of the athlete has been improved during plyometric training in the form of jump‐landing training, which increases the dynamic stability in landing.

Current research evidence indicates that loads on the ACL may not originate from a single plane but rather from the combined loads of multiple planes (Markolf et al., 1995; Oh et al., 2012). For example, multiple video analyses of ACL injuries indicate that concomitant smaller flexion angles, larger knee valgus angles and internal rotation angles of the knee are often present at the moment of injury (Olsen et al., 2004; Shimokochi et al., 2008). Meanwhile, the combined loading pattern of knee internal rotation, extension and abduction moments has been considered to place the greatest load on the ACL (Markolf et al., 1995; Oh et al., 2012). Collectively, these combined biomechanical characteristics contribute to the appearance of ACL injuries. Results have shown that athletes receiving plyometric training presented higher knee flexion angles, lower valgus angles and lower three‐dimensional moments in landings. This outcome indicates that the combined load on the three planes of knee in landing is reduced and the pressure bearing on the ACL is consequently reduced, thus decreasing the risk of injury. Meanwhile, lower loading rates on the plyometric training group are part of the evidence for a reduced risk of injury. The loading rate reflects the ground reaction force borne by the body per unit of time, and the decrease in its value partially responds to the reduction in stress on the knee (Abolins et al., 2019). It has been stated that impact loading of the knee joint is one of the reasons why acute or overuse injuries of the lower extremity occur in recreational and elite sports athletes (McManus et al., 2006). Therefore, strategies to reduce landing impact may reduce the risk of ACL injury (Lee et al., 2021).

The possibility exists that these biomechanical changes are related to altered neuromuscular adaptations and differences in motor control resulting from different forms of training. Plyometric training differs from traditional strength in that plyometric training attempts to utilise the stretch shortening cycle, which may enable to invoke particular neuromotor adaptations, such as an increased activation of motor units (Mikkola et al., 2007). Such changes in neuromuscular adaptations during landing may be able to inhibit dangerous motion patterns in the lower limbs that would otherwise result from deficits in neural control such as excessive external or internal rotation of the knee. Lloyd et al. (2001) showed in their study that after 6 weeks of plyometric training subjects presented more significant EMG signals in the hip abductors and adductors during the early stages of landing, which suggested that the neuromuscular system can adapt to facilitate more rapid and effective stabilisation during landing, thereby enhancing joint stability. Besides, plyometric training is filled with numerous jumps and landings, and subjects may have learnt how to increase knee motor control during landings in the process. For example, it has been indicated that the medial and lateral hamstrings are selectively activated to control tibial internal and external rotation and that activation of the lateral hamstrings may be a critical factor in preventing rotation during the pre‐landing phase (Besier et al., 2003; Ciccotti et al., 1994). Whereas, subjects may have also learned to activate the medial hamstrings less to maximise the lateral hamstrings to stabilise the knee against internal/external rotational moments when they practised landing in plyometric training repetitions (Lephart et al., 2005). In addition, Viitasalo et al. (1998) compared EMG of the lower limbs of subjects receiving plyometric training and controls during landing and found that the plyometric training group showed more intense lateral hamstring activation during the pre‐landing phase.

Furthermore, the results of the present study indicated an interaction effect between the type of training and duration of training for changes in the landing biomechanical characteristics of athletes. For instance, no significant differences were observed in the knee flexion angle, internal rotation angle, extension moment and internal rotation moment between the plyometric training group and the traditional training group at week 8 of training, whereas these biomechanical characteristics presented a significant difference at week 16. Current research on plyometric training focuses on its short‐ and medium‐term effects (4–8 weeks), and some discrepancy remains between existing studies. (Myer, Ford, McLean, & Hewett, 2006) observed that plyometric training increased the knee flexion angle and decreased the valgus angle. Meanwhile, Brown et al. (Tsang et al., 2011) indicated that plyometric training increased the knee flexion angle in landing with no change in the valgus angle. We consider that this discrepancy can be mainly attributed to the difference in training time. On the one hand, short‐term plyometric training may not be effective in showing significant changes because of individuals' differences in adaptation to the training. On the other hand, a meta‐analysis by Wang et al. (2023) indicated that longer‐term plyometric training contributes to lower‐limb muscle strength and that the duration of training may significantly affect muscle strength improvement, thus resulting in differences in landing biomechanical characteristics. The results of our study further support the idea that longer training time has a positive impact on reducing the risk of ACL injury.

Some limitations remain in this study. Firstly, plyometric training cannot be aligned with traditional strength training in terms of intensity division. The intensity of plyometric training, which we have opted to develop based on the guideline protocols, cannot be individualised in the same way as the traditional training group. Hence, the intensity of the exercise performed by the two groups in the actual training may differ. This discrepancy has the potential to affect the final assessment results. Additional insightful and detailed studies on the intensity of plyometric training should be conducted. Secondly, the plyometric training program has certain weaknesses. No other form of training was involved in the plyometric training program used in this study. Notably, few plyometric training programmes include strength training for the hamstrings. Although hamstring strength is also critical in the prevention of ACL injury risk, the present study did not explore whether plyometric training could be applied to hamstring strengthening and the associated biomechanical changes. Finally, the study was not blinded to the experiment performers and subjects, which might affect the risk of bias in this study partially.

5. CONCLUSION

Compared with traditional training, 16 weeks of plyometric training improved hip extension and flexion strength as well as knee extension strength in collegiate basketball players. However, it had no significant effect on the improvement of flexion strength. The flexion angle significantly increased, and the internal rotation angle, loading rate and maximum knee extension moment significantly decreased in the plyometric training group at week 8 of training. Then, at week 16 of training, the plyometric training group showed an increase in knee flexion angle, a decrease in rotation angle and a decrease in load rate, extension and abduction moments at the peak moment of vGRF, maximum extension moment and rotation moment. Notably, 8 and 16 weeks of plyometric training lead to changes in biomechanics associated with reduced risk of ACL injury, and the effect was more pronounced with long‐term training.

AUTHOR CONTRIBUTIONS

Bocheng Chen wrote the main manuscript text and performed data analysis; Ziyan Ye conduct experimental designs and training plans; Jiaxin Wu and Tiancheng Yu assistance with data collection; Guoxiang Wang captured research ideas and provided writing guidance.

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no potential conflicts of interest with respect to the research, author‐ship and/or publication of this article. The authors wish to assert that there are no conflicts of interest associated with any of the authors or organisations involved in this study. The author declare that they have no financial or personal relationships that could influence the findings or interpretation presented in this paper.

PATIENT CONSENT STATEMENT

All participants provided informed consent.

DATA AVAILABILITY STATEMENT

The authors declare the data that support the findings of this study are available on request from the corresponding author upon reasonable request.