Estimation and Comparison of Knee Joint Contact Forces During Heel Contact and Heel Rise Deep Squatting

Abstract

Background

Increased knee flexion is required for deep squatting in the daily life of the non-western population as well as in many sports activities. The purpose of this study was to estimate as well as to compare knee joint contact forces during heel contact (HC) and heel rise (HR) deep squatting in 10 healthy young Indian participants.

Materials and Methods

Kinematic data were captured using a 12-camera Motion Analysis system. Kinetic data were collected using two Kistler force plates. EMG of 6 lower limb muscles was monitored by Noraxon wireless EMG. OpenSim musculoskeletal model was customized to increase the maximum knee flexion capability of the existing model and knee joint contact forces were estimated.

Results

A significant difference in tibiofemoral (p < 0.001) as well as patellofemoral (p = 0.006) knee joint contact force was observed between HC and HR squatting. The resultant maximum tibiofemoral KJCF was 5.9 (± 0.54) times body weight (BW) and 5.3 (± 0.6) BW for the HC and HR, respectively. The resultant maximum patellofemoral KJCF was 7.8 (± 0.57) BW and 7.1 (± 0.73) BW for the HC and HR, respectively.

Conclusion

The findings can provide implications for physiotherapists to design rehabilitation exercise protocols, exercise professionals, and the development of high flexion knee implants.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s43465-022-00798-y.

Introduction

The knee joint is a highly complex and heavily loaded joint in the human body. Knowledge of knee joint contact forces is important for different treatment options and for optimizing clinical outcomes [1]. Longer cumulative exposure to squatting leads to a higher risk of osteoarthritis of the knee [2]. Deep squatting is very common all over the world especially in non-western countries while performing the activity of daily living. Deep squatting can also be seen in many sports activities like football, basketball, baseball, and weight lifting. Baseball/softball catchers where squatting is repeatedly performed develop osteochondritis dissecans [3]. Athletes are recommended to perform parallel squats (0° to 100° knee flexion) over deep squats as parallel squats are not injurious to a healthy knee [4]. Consequently, an understanding of mechanical loading is very important during deep squatting.

Peak Tibiofemoral (TF) contact forces are much larger in squatting than in normal walking [5–8]. Three main techniques to determine knee joint contact forces (KJCF) are mathematical modeling, direct experimental measurement in vivo using telemetry, and using a cadaver in vitro. A recent systematic review [9] showed that TF contact forces estimated theoretically range from 2.3 to 9 times body weight (BW). Whereas peak TF contact forces measured in vivo experimentally range from 2.1 to 3 BW but peak flexion angle achieved during squatting was from 71° to 104° only. Peak Patellofemoral (PF) contact forces were estimated theoretically range from 3.2 to 7.5 BW. No study measured in vivo PF force experimentally. PF KJCF measured experimentally by vito cadaver studies ranges from 4 to 9 BW. Knee joint contact force (KJCF) during squatting predicted or measured by using different techniques varies significantly.

Nowadays musculoskeletal modeling and simulation software available show more accurate predictions than two-dimensional mathematical modeling. Several musculoskeletal modeling software like OpenSim, Anybody, LifeModeler, MSIM, BodyMech, etc. are available and can be used to estimate knee load. OpenSim [10] is an open-source musculoskeletal modeling and simulation software that estimates the internal loading of the musculoskeletal system. A recent study [11], implemented new and updated the existing muscle wrapping surfaces of the Lai Model [12] to make the model suitable for analysis up to 138° hip and 145° knee flexions. To better predict KJCF for tasks requiring a large range of motion at the knee, the musculoskeletal models (MSKM) need appropriate properties of muscle tendons and muscle paths. Appropriate moment arm (MA) lengths are essential for muscle paths crossing the bone.

Heel contact and heel rise deep squatting are the most common postures in non-western countries. In countries like India, China, Japan, and Arab countries, the activity of daily living includes deep squatting and the maximum knee flexion reported is 165° [13]. In recent years, there has been an increasing demand for accurate disease diagnosis and individualized therapy. The musculoskeletal modeling and simulation software can be used in series with the digital twin of a subject-specific or geographical model of the knee joint generated using CT and MRI to simulate deep squatting. Techniques like finite element analysis (FEA) may give a clear understanding of the causes of knee problems, possible treatments, and injury prevention. KJCF in the typical deep squat posture has been studied least, with only three previous studies [14–16] quantified KJCF in the Asian population. To the author's knowledge, this study is the first study quantifying KJCF during deep squatting in the Indian population.

The purpose of the current study was to estimate and compare KJCF during heel contact and heel rise deep squatting. We modified the existing OpenSim musculoskeletal model to increase knee range of motion capability and estimated KJCF using a modified model. We also qualitatively compared muscle activations predicted by a model with measured EMG.

Methods

Participants and experimental protocol

Ten healthy young Indian participants (5 male, 5 female) participated with a mean standard deviation age of 27.8 (± 5.71) years, a height of 1.66 (± 0.083) m, and a weight of 58.14 (± 7.8) kg. According to the declaration of Helsinki's ethical guidelines for medical research involving human participants, each participant read and signed an informed consent form. Kinematic data were captured using twelve-camera Motion Analysis (Motion Analysis Corporation, Santa Rosa, CA, USA) with Cortex 5.0 software at a frequency of 100 Hz. Kinetic data were collected using two Kistler force plates at a frequency of 1000 Hz. EMG of 6 lower limb muscles vastus medialis, vastus lateralis, lateral hamstring, medial hamstring, gastrocnemius, and tibialis anterior was monitored by Noraxon (Noraxon USA, Inc., Scottsdale, AZ, USA) wireless surface EMG at 1000 Hz on the of the dominant (right) leg. A previously established Helen-Hayes 29 marker set was used to trace the motion of the body.

Before the experiment, participants were instructed to perform practice squatting movements. The squatting movement started with each foot stepping on the force platform. Participants were instructed to perform deep squats as low as possible with a) heel contact and b) heel raise posture as shown (Fig. 1) with their self-sustained movement pattern (Supplemental Fig. 2).

Fig. 1.

OpenSim model is shown for an exemplary subject during a HC b HR squatting and c coordinate system for tibia and patella

Fig. 2.

Workflow of the study

Motion capture data and ground reaction forces were filtered using a fourth-order Butterworth low-pass filter with a cut-off frequency of 6 Hz using MATLAB script [17] EMG signals were band-pass filtered between 10 and 300 Hz, rectified, and low-pass filtered at 6 Hz using 4th order Butterworth filter using MATLAB scripts [17]. EMG data were normalized to the maximum muscle activation value across all trials.

Musculoskeletal model modification

OpenSim 4.2 was used to perform simulations. To allow simulation-based studies of deep squatting involving high knee flexions, we increased maximum knee flexion from 145° to 160° in the existing model [11]. Four additional fixed muscle attachment points were added to the rectus femoris, vastus intermedius, vastus medialis and vastus lateralis (Supplemental Table 1). Two wrapping surfaces (WS) of gastrocnemius lateralis and gastrocnemius medialis were updated (Supplemental Table 2). By adding an attachment point, we forced the muscle–tendon unit to respect its path over the existing WS during deep flexion. Simulations were performed for the original and modified MSKM. Motion capture data of a participant who achieved maximum knee flexion was used to validate modifications. The MSKM was first scaled based on a static pose. The inverse kinematics (IK) tool was used to compute joint kinematics. Muscle activations were calculated using static optimization (SO) while minimizing the sum of squared muscle activations. Muscle analysis was used to calculate the moment arm (MA) of the original and modified model (Fig. 2). The MA lengths of the vastus intermedius, vastus lateralis, vastus medialis, rectus femoris, gastrocnemius lateralis, and gastrocnemius medialis were calculated during squat for original as well as modified MSKM. Calculated MA lengths of the original and modified model were compared to experimental data [18]. Muscle activations of original and modified MSKM were also compared with measured EMG.

Musculoskeletal model simulation to predict KJCF

The modified model was then scaled for each participant using skin markers placed on the bony landmark. IK analysis was then performed to calculate the joint angles. SO tool which minimizes the sum of the muscle activations squared at each time used to determine muscle activation and forces. Joint Reaction analysis (JRA) was then performed to compute KJCF (Fig. (Fig. 2) at the knee in the superior (Y), anterior (X), and medial (Z) direction (tibial reference frame for TF and patella reference frame for PF). We also qualitatively compared the computed muscle activity to the participant’s recorded EMG data over the percent squat cycle. Guidelines [19] provided were used for simulations. To make information general and less subject-specific, all forces were normalized to BW.

Statistical Analysis

1D-Statistical parametric mapping (1dSPM) was used for statistical analysis [20]. The paired sample t-test package in the 1dSPM was used with a significance level of p < 0.05 to detect significant differences between HC and HR squatting. Data analysis was performed using MATLAB R2021a.

Results

Modified model

No differences in kinematics were observed between the original and modified models. Visual inspection of the muscle paths shows that modifications prevented knee muscles from crossing the bony structures during the deep squatting task (Supplemental Fig. 1). Modifying the two knees WS (GasMed_at_shank, GasLat_at_shank) improved the MA of the gastrocnemius for knee flexion angles larger than 145°. Adding fixed attachment points to quadriceps muscles also improved MA for knee flexion angles larger than 145° (Fig. 3). Simulated muscle activations of a participant were compared with measured EMG signals (Fig. 4) for both original and modified models. Modified model muscle activation avoided abnormal changes.

Fig. 3.

Model moment arms of the modified muscles compared to the original model and experimental data

Fig. 4.

Predicted muscle activations of the modified model compared to the original model and measured EMG

KJCF using a modified model

There was no significant difference between knee flexion/extension angle during heel contact and heel rise squatting (Supplemental Fig. 3). TF KJCF showed a significant difference (p < 0.001) between heel contact and heel rise squatting in the superior directions. No significant difference in TF KJCF was observed in the medial and anterior directions (Fig. 5). PF KJCF showed a significant difference (p = 0.006) between heel contact and heel rise squatting in the anterior directions. No significant difference in PF KJCF was observed in the medial and superior directions (Fig. 5). The maximum achieved knee flexion angle by a participant was approximately 160° for heel contact and heel rise squatting.

Fig. 5.

Predicted a TF and b PF KJCF during heel contact and heel rise squatting for all subjects and their comparison using 1dSPM

Muscle activation comparison with measured EMG

There was good agreement between the trends in predicted muscle activations and the EMG measurements for vastus lateralis, vastus medialis, medial hamstring, and gastrocnemius. However, there were differences in predicted muscle activity and EMG of the lateral hamstring and tibialis anterior for both HC and HR (Fig. 6) squatting.

Fig. 6.

The predicted muscle activations of the modified model for all subjects compared to measured EMG for a HC and b HR squatting

Discussion

We aimed to estimate KJCF during heel contact and heel rise deep squatting. The mean peak knee flexion angle during HC squatting was 150.8 (± 6.4) ° and 151.5 (± 6.2) ° during HR squatting indicating the type of squatting does not affect the depth of the squat. The musculoskeletal model was modified to allow a deeper knee flexion angle. The maximum allowable hip flexion angle of an existing model (138°) was suitable for our simulations. The results indicate that there is a significant difference in KJCF between HC and HR squatting during the descent phase. The HC squat yielded a higher mean TF contact force than the HR squat. KJCF were large during descent (around 0–40% squat cycle) and ascent (around 60–100% squat cycle) and were less in rest position (around 40–60% squat cycle) in a full squat (Fig. 5).

The resultant maximum TF KJCF was 5.9 (± 0.54) BW for descent and 5.6 (± 0.43) BW for ascent for HC squat. However, for HR squat resultant maximum TF KJCF was 5.3 (± 0.6) BW for descent and 4.9 (± 0.5) BW for the ascent. The estimated TF KJCF in this study was comparable with previous deep squatting studies using OpenSim studies [16, 21]. The resultant maximum PF KJCF was 7.7 (± 0.57) BW for descent and 7.8 (± 0.71) BW for ascent for HC squat. However, for HR squat mean PF KJCF was 7.1 (± 0.73) BW for descent and 6.8 (± 1) BW for the ascent. The estimated TF and PF KJCF are well aligned with the ones provided in the literature [9, 22]. Although squatting is good for muscle development, it causes very high mechanical loads in the knee increasing the chances of damaged articular cartilage.

To the author's knowledge, only one study [23] determines KJCF during both HC and HR squatting where estimated compressive contact forces for HR squatting (3.8 BW) are more than HC squatting (2.8 BW). This was due to the frequent use of railing by the Canadian subjects living the western lifestyle for stability during HC squatting. None of our participants use support during both types of squatting. According to the systematic review [9], the mean peak TF contact force in vivo experimental studies was 2.5 BW (± 0.3) and all subjects were Caucasian. The limitation of in vivo experimental study was peak knee flexion angle achieved was only 85.2° (± 9.7°). The mean peak TF and PF contact force by theoretical studies of Caucasian subjects were 5.8 (± 2.1) and 5.3 (± 2) respectively. The mean peak TF contact force by theoretical studies of Asian subjects was 3.5 (± 1.2). Differences in KJCF estimated in this study with other ethnicity studies may be due to factors like squatting stance width, the orientation of the foot, the speed at which squatting was performed by Indian participants, and the model used. Further research is demanded to understand the reasons for these differences.

It was observed that KJCF increased up to around 115° and then decreased till maximum knee flexion in both types of the squat. A similar trend was observed in previous OpenSim studies [16, 21] but the maximum knee flexion angle during these studies was only around 145°. The anterior contact force was very high in both types of squatting. Shear force i.e. anterior force is very important as it is responsible for loosening the implant [24]. Moreover shear may be responsible for a destructive profile of cartilage [25, 26]. Thigh calf contact force was not considered in this study. However, the present study emphasizes peak contact forces where thigh-calf contact force is no or minimal. Thigh calf contact starts at 128° and 125° for HC and HR squat, respectively [27, 28] but peak KJCF was found well below this angle.

HR squat showed more gastrocnemius muscle EMG activity than the HC squat. HC squat shows more tibialis anterior EMG activity than the HR squat (Fig. 6). High quadriceps muscle activity was observed around 115° knee flexion not at the deepest squat suggesting squatting past this angle might not be useful for muscle development. Similar results were observed in a previous study [29] where muscle activity was greater at around 90° but not at the deepest squat. Heel contact squat exerts a more anterior force on the knee and hence a heel rise with a wedge could be more beneficial to avoid implant loosening post-operatively also the heel rise squatting could avoid patellofemoral pain or early cartilage degeneration in non-TKR patients. Also, our study provides information that the two squat types did not show any significant difference in the knee range of motion (ROM) (Supplemental Fig. 3.). This information will help the physiotherapist to plan the knee ROM passive and active and accordingly work on mobilization.

Recently, high-flexion knee implants have gained popularity. Although high-flexion knee implants allow deeper flexion, does not demonstrate any benefits and even increase the likelihood of implant loosening using the mobile-bearing high-flex knee [30]. FEA can be used to predict implant biomechanical behavior, wear, stress, pressure, and laxity under different loading conditions. Musculoskeletal modeling on the other hand can predict patient-specific biomechanics by including muscle force and joint contact forces. Predicted muscle forces or joint contact forces can be used in series with FEA of knee implants to understand the mechanical reasons for implant failures. Predicted joint contact forces can also be used as input for the FEA of a digital twin of the knee to understand tissue-level mechanics such as contact pressure, and stress because stress higher than normal on the articular cartilage may cause knee osteoarthritis. Cadaver studies are very useful to determine knee joint contact area and pressure using a knee joint simulator. Knee joint contact forces can also be used as input to the knee joint simulator.

There are some limitations of this study. Default parameters were used during simulations on OpenSim. Scaling and IK were performed using a marker-based which can introduce significant measurement errors. The OpenSim model used in this study included only one degree of freedom at the knee and ankle. Estimated muscle activations of the lateral hamstring and tibialis anterior are not consistent with EMG. Further research and different methods are needed to investigate this difference.

Conclusion

Greater KJCF and quadriceps muscle activation were reported during both types of squatting which peaks at around 115° and then decreases till maximum flexion. In this study, we also observed a significant difference in KJCF between heel contact and heel rise squatting during the descent phase but there was no difference in the deepest position. Future studies should quantify the KJCF using a modified model incorporating thigh-calf contact force to understand the effect of a sustained deep squat. Results from the current study may provide implications for physiotherapists while designing protocols and exercise professionals for the prevention of loading accumulation during squatting. KJCF predicted can be used as input to finite element models to estimate detailed tissue mechanics that predict cartilage degeneration. The resulting data can also be used as input implant design during finite element analysis or joint simulator testing in populations where deep squatting is the predominant activity of daily living.

Supplementary Information

Below is the link to the electronic supplementary material.

Declarations

Conflict of interest

No potential conflict of interest was reported by the authors.

Ethical standard statement

This article does not contain any studies with human or animal subjects performed by the any of the authors

Informed consent

For this type of study informed consent is not required.